Learning spatiotemporal dynamics from sparse data via a high-order physics-encoded network

Pu Ren, Jialin Song, Chengping Rao, Qi Wang, Yike Guo, Hao Sun, Yang Liu

Abstract: Highlights•A high-order physics-encoded learning framework is proposed for learning spatiotemporal dynamics.•This paper focuses on coefficient identification and high-resolution dynamics reconstruction.•The known physical laws are encoded into the network and a high-order time marching scheme is considered.•The presented method will contribute to the advancement of the physics-aware machine learning community.