An LSQR-based algorithm for large-scale null space computations

DOI10.48550/arXiv.2607.03341MaRDI QIDQ7283872zbMATH OpenarXivFDO

Authors Jinzhi Huang

Publication date 3 July 2026

Published in arXiv

Tldr We propose an LSQR-based algorithm, LSQRNV, to compute a null vector of a large rank-deficient matrix from an initial vector and extend it to LSQRNS to obtain an orthonormal basis of the null space with a rigorous accuracy bound; further, we can check if a matrix is numerically rank-deficient.

Full work available at URL https://doi.org/10.48550/arXiv.2607.03341

Mathematics Subject Classification ID

Vector spaces, linear dependence, rank, lineability (15A03) Linear equations (linear algebraic aspects) (15A06) Iterative numerical methods for linear systems (65F10)

arXiv classification

math.NA

This page was built for publication: An LSQR-based algorithm for large-scale null space computations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7283872)