Id: arxiv/2606.29083 Url: https://arxiv.org/abs/2606.29083 Tags: math.NA, cs.LG, math.DS Score: 0 Title: Residual-Guided Dictionary Learning for Spectrally Accurate Koopman Approximation Domain: math.NA Source: arxiv Authors: George Coote, Matthew J. Colbrook Reviews: 0 Subject: math.NA Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-06-27 Page: https://delphi.science/r/arxiv/2606.29083 Koopman theory promises linear structure in nonlinear dynamics, but numerical Koopman spectra are easy to compute and hard to trust. A finite EDMD matrix always has eigenvalues; the problem is that many of them may have nothing to do with the infinite-dimensional operator. In this paper we make spectral reliability the objective of dictionary learning. We train neural-network dictionaries not merely to predict the next snapshot, but to minimize Residual Dynamic Mode Decomposition residuals: operator-level a posteriori errors that test whether computed eigenvalues and modes are genuine…