Id: arxiv/2611.11257 Url: https://arxiv.org/abs/2611.11257 Tags: math.DS, cs.LG Score: 0 Title: Learning calibrated probabilistic forecasts using Koopman mode decomposition and the continuous ranked probability score Domain: math.DS Source: arxiv Authors: Benjamin M. Kent, Richard E. Turner Reviews: 0 Subject: math.DS Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-11-08 Page: https://delphi.science/r/arxiv/2611.11257 Well-calibrated forecasting for non-linear dynamical systems is essential in science and engineering applications such as fluid dynamics and weather prediction. We present a forecasting method combining probabilistic machine learning with the structured linear dynamics of Koopman mode decomposition. Within this framework, the observation, linear dynamics, and state reconstruction operators are constructed using standard machine learning architectures, producing predictive distributions via the pushforward of injected noise samples. The model is trained using a proper scoring rule, the…