Stochastic trace estimation with tensor train random vectors

DOI10.48550/arXiv.2606.15679MaRDI QIDQ7270000arXivFDO

Authors Zvonimir Bujanović, Daniel Kressner, Hrvoje Olić

Publication date 14 June 2026

Published in arXiv

Tldr We introduce Gaussian random tensor train vectors for stochastic trace estimation and show they achieve dimension-independent guarantees, matching those of unstructured vectors with a suitable rank choice. We also prove an oblivious subspace injection result and demonstrate their use within the Nyström++ framework.

Full work available at URL https://doi.org/10.48550/arXiv.2606.15679

arXiv classification

cs.LGmath.NAstat.ML

This page was built for publication: Stochastic trace estimation with tensor train random vectors

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7270000)