Finite-time Cholesky Factorization

DOI10.48550/ARXIV.2611.09706MaRDI QIDQ7422371arXivFDO

Authors Rahul Manavalan, Carola Bibiane Schoenlieb

Publication date 6 November 2026

Published in arXiv

Tldr We propose a finite-time algorithm for estimating the Cholesky factorization of positive, semi-definite matrices without pivoting by interpreting the factorization task as a variational problem and solving it using an induced dynamical system. Our method has the same data access pattern as the canonical Cholesky method and is amenable to familiar hardware optimization techniques.

Full work available at URL https://doi.org/10.48550/ARXIV.2611.09706

arXiv classification

math.NA

Cites work

• A mixed precision LOBPCG algorithm
• A set of level 3 basic linear algebra subprograms
• LAPACK Users' Guide
• Mixed precision algorithms in numerical linear algebra
• NOTES ON THE SOLUTION OF ALGEBRAIC LINEAR SIMULTANEOUS EQUATIONS
• Randomly pivoted Cholesky: practical approximation of a kernel matrix with few entry evaluations
• Robust, randomized preconditioning for kernel ridge regression
• ROUNDING-OFF ERRORS IN MATRIX PROCESSES
• Sparse Cholesky factorization for solving nonlinear PDEs via Gaussian processes

This page was built for publication: Finite-time Cholesky Factorization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7422371)