{"id":"arxiv/2609.14812","url":"https://arxiv.org/abs/2609.14812","tags":["math.NA","physics.flu-dyn"],"score":0,"title":"Learning Long-Term Stable Operator Inference Reduced-Order Models of Fluid Flows through Online Spatial Filtering","domain":"math.NA","source":"arxiv","authors":["Ian Moore","Ping-Hsuan Tsai","Anthony Gruber","Ionuţ Farcaş","Christopher Wentland","Irina Tezaur","Traian Iliescu"],"reviews":0,"subject":"math.NA","summary":"This paper introduces an online evolve--filter--relax (EFR) strategy for long-term stability of Operator Inference (OpInf) reduced-order models (ROMs) of complex fluid flow simulations. The main novelty of the new EFR-OpInf strategy is the use of online (i.e., at the learned ROM online evaluation level) spatial filtering inspired from large eddy simulation to significantly improve long-term stability and predictive performance of standard OpInf. Furthermore, the EFR-OpInf strategy reduces, and in some cases even eliminates, the need for standard $L^2$ regularization, while providing physical…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-13","reactions":{}}