Id: arxiv/2603.15091 Url: https://arxiv.org/abs/2603.15091 Tags: math.NA, cs.LG, math.DS, math.OC Score: 0 Title: Trustworthy Koopman Operator Learning: Invariance Diagnostics and Error Bounds Domain: math.NA Source: arxiv Authors: Gustav Conradie, Nicolas Boullé, Jean-Christophe Loiseau, Steven L. Brunton, Matthew J. Colbrook Reviews: 0 Subject: math.NA Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-16 Page: https://delphi.science/r/arxiv/2603.15091 Koopman operator theory provides a global linear representation of nonlinear dynamics and underpins many data-driven methods. In practice, however, finite-dimensional feature spaces induced by a user-chosen dictionary are rarely invariant, so closure failures and projection errors lead to spurious eigenvalues, misleading Koopman modes, and overconfident forecasts. This paper addresses a central validation problem in data-driven Koopman methods: how to quantify invariance and projection errors for an arbitrary feature space using only snapshot data, and how to use these diagnostics to produce…