data-driven simulation is an important step-forward in computational physics
when traditional numerical methods meet their limits. Learning-based simulators
have been widely studied in past years; however, most p
ConvSSMs combine ConvLSTM and state space methods to efficiently model long spatiotemporal sequences, outperforming Transformers and ConvLSTM in terms of training speed and sample generation while matching or exceeding state-of-the-art methods on various benchmarks.