Learning elliptic partial differential equations with randomized linear algebra

DOI10.1007/S10208-022-09556-WMaRDI QIDQ6359476arXivWikidataFDO

Authors Nicolas Boullé, Alex Townsend

Publication date 31 January 2021

Published in arXiv

Full work available at URL https://doi.org/10.1007/S10208-022-09556-W

Mathematics Subject Classification ID

Green's functions for elliptic equations (35J08) Gaussian processes (60G15) Numerical solutions to overdetermined systems, pseudoinverses (65F20) Numerical methods for low-rank matrix approximation; matrix compression (65F55) Fundamental solutions, Green's function methods, etc. for boundary value problems involving PDEs (65N80)

arXiv classification

cs.LGcs.NAmath.NA

Abstract: Given input-output pairs of an elliptic partial differential equation (PDE) in three dimensions, we derive the first theoretically-rigorous scheme for learning the associated Green's function $G$. By exploiting the hierarchical low-rank structure of $G$, we show that one can construct an approximant to $G$ that converges almost surely and achieves a relative error of $mathcalO(Gamm{a}_{e}psilo{n}^{-1/2}lo{g}^3(1/epsilon)epsilon)$ using at most $mathcalO(epsilo{n}^{-6}lo{g}^4(1/epsilon))$ input-output training pairs with high probability, for any $0<epsilon<1$. The quantity $0<Gamm{a}_{e}psilonleq1$ characterizes the quality of the training dataset. Along the way, we extend the randomized singular value decomposition algorithm for learning matrices to Hilbert--Schmidt operators and characterize the quality of covariance kernels for PDE learning.

This page was built for publication: Learning elliptic partial differential equations with randomized linear algebra

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6359476)