{"id":"arxiv/2603.28074","url":"https://arxiv.org/abs/2603.28074","tags":["cs.LG","math.DS"],"score":0,"title":"Koopman-based surrogate modeling for reinforcement-learning-control of Rayleigh-Benard convection","domain":"cs.LG","source":"arxiv","authors":["Tim Plotzki","Sebastian Peitz"],"reviews":0,"subject":"cs.LG","summary":"Training reinforcement learning (RL) agents to control fluid dynamics systems is computationally expensive due to the high cost of direct numerical simulations (DNS) of the governing equations. Surrogate models offer a promising alternative by approximating the dynamics at a fraction of the computational cost, but their feasibility as training environments for RL is limited by distribution shifts, as policies induce state distributions not covered by the surrogate training data. In this work, we investigate the use of Linear Recurrent Autoencoder Networks (LRANs) for accelerating RL-based…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-03-30","reactions":{}}