Near-Optimal Learning of Gaussian Sobolev Operators

DOI10.48550/arXiv.2607.11921MaRDI QIDQ7284035arXivFDO

Authors Ben Adcock, Michael Griebel, Gregor Maier

Publication date 8 July 2026

Published in arXiv

Tldr We present an algorithm, Hermite-PCA approximation, that learns Gaussian Sobolev operators efficiently by combining principal component analysis and weighted least-squares methods, achieving near-optimal sample complexity rates for a wide range of regularities.

Full work available at URL https://doi.org/10.48550/arXiv.2607.11921

arXiv classification

cs.ITcs.LGmath.NA

This page was built for publication: Near-Optimal Learning of Gaussian Sobolev Operators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284035)