Randomized Block Davidson Eigensolvers for Plane-Wave Density-Functional Theory

DOI10.48550/arXiv.2608.24529MaRDI QIDQ7285073arXivFDO

Authors Moritz Gubler, Nicola Marzari, Iurii Timrov, Laura Grigori, Taejun Park, Augustin Bussy, Johannes Laute

Publication date 25 August 2026

Published in arXiv

Tldr We present a randomized block Davidson-type eigensolver that replaces Euclidean orthogonalization with randomized Gram-Schmidt and applies it to plane-wave density-functional theory. This modification changes only the Rayleigh-Ritz step into a definite generalized Hermitian eigenproblem, while preserving true Ritz pairs and interlacing property, outperforming deterministic counterparts beyond $2\times 10^4$ matrix dimensions.

Full work available at URL https://doi.org/10.48550/arXiv.2608.24529

arXiv classification

cond-mat.mtrl-scimath.NAphysics.comp-ph

Cites work

• A Jacobi--Davidson Iteration Method for Linear Eigenvalue Problems
• Fast and accurate randomized algorithms for linear systems and eigenvalue problems
• Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions
• Generalizations of Davidson’s Method for Computing Eigenvalues of Sparse Symmetric Matrices
• Improved analysis of the subsampled randomized Hadamard transform
• Improved approximation algorithms for large matrices via random projections
• Low-rank approximation and regression in input sparsity time
• Nearly tight oblivious subspace embeddings by trace inequalities
• Numerical methods for large eigenvalue problems
• OSNAP: faster numerical linear algebra algorithms via sparser subspace embeddings
• PRIMME: preconditioned iterative multimethod eigensolver -- methods and software description
• Randomized block Gram-Schmidt process for the solution of linear systems and eigenvalue problems
• Randomized Gram-Schmidt process with application to GMRES
• Randomized Numerical Linear Algebra: Foundations & Algorithms
• Rounding error analysis of the classical Gram-Schmidt orthogonalization process
• Sketching as a tool for numerical linear algebra
• The Davidson Method
• The fast Johnson-Lindenstrauss transform and approximate nearest neighbors
• The iterative calculation of a few of the lowest eigenvalues and corresponding eigenvectors of large real-symmetric matrices
• Toward the optimal preconditioned eigensolver: Locally optimal block preconditioned conjugate gradient method

This page was built for publication: Randomized Block Davidson Eigensolvers for Plane-Wave Density-Functional Theory

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7285073)