Accelerating preconditioned Jacobi methods via perturbation-inspired pivoting

DOI10.48550/arXiv.2607.23187MaRDI QIDQ7284436arXivFDO

Authors Nian Shao, Yuji Nakatsukasa

Publication date 25 July 2026

Published in arXiv

Tldr This paper accelerates the Jacobi method for symmetric eigenvalue problems by incorporating spectral gap information into pivoting and combining it with a mixed-precision preconditioner. The new strategy significantly outperforms the traditional greedy approach, especially for matrices with clustered eigenvalues.

Full work available at URL https://doi.org/10.48550/arXiv.2607.23187

arXiv classification

math.NA

Cites work

• A mixed precision Jacobi SVD algorithm
• A note on a one-sided Jacobi algorithm
• A note on eigenvalues of perturbed Hermitian matrices
• Accuracy and Stability of Numerical Algorithms
• Accurate Eigensystem Computations by Jacobi Methods
• Bounds on the singular values of matrices with displacement structure
• Computing accurate eigenvalues using a mixed-precision Jacobi algorithm
• Computing accurate eigenvalues using a mixed-precision Jacobi algorithm
• Computing accurate singular values using a mixed-precision one-sided Jacobi algorithm
• Inversion of Matrices by Biorthogonalization and Related Results
• Jacobi’s Method is More Accurate than QR
• Minimizing the Arithmetic and Communication Complexity of Jacobi's Method for Eigenvalues and Singular Values
• Mixed precision algorithms in numerical linear algebra
• New Fast and Accurate Jacobi SVD Algorithm. I
• New Fast and Accurate Jacobi SVD Algorithm. II
• Note on the quadratic convergence of the cyclic Jacobi process
• Über ein leichtes Verfahren die in der Theorie der Säcularstörungen vorkommenden Gleichungen numerisch aufzulösen*).

This page was built for publication: Accelerating preconditioned Jacobi methods via perturbation-inspired pivoting

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284436)