Convergence Rates for Learning Pseudo-Differential Operators

Jiaheng Chen and Daniel Sanz-Alonso · 2026

Read Convergence Rates for Learning Pseudo-Differential Operators by Jiaheng Chen and Daniel Sanz-Alonso (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-04473.txt.

Description

Jiaheng Chen and Daniel Sanz-Alonso put error bars on learning the laws of physics themselves, in a 2026 arXiv paper. Their target is a basic class of operators from partial differential equations and physics, the elliptic pseudo-differential operators, which turn inputs into outputs across many scales at once. Learning such an operator from data is an infinite-dimensional regression, and theory for how fast the learning converges has lagged the algorithms.

In a wavelet-Galerkin scheme, the authors exploit the class's multiscale sparsity with a new matrix compression scheme and a nested-support strategy that balance approximation against estimation error. They prove convergence rates for the resulting sparse estimator, then show the learned operator doubles as a working solver whose own error matches its statistical accuracy. Operator learning, data-driven solvers, and wavelet methods, usually separate literatures, meet in the argument.

Catalog record

CREATORJiaheng Chen and Daniel Sanz-Alonso
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE260 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-04473

Browse the full library · About the Tower of Babel