[01151] Structure-Preserving Neural Networks for Hamiltonian Systems
Session Time & Room : 3D (Aug.23, 15:30-17:10) @E606
Type : Contributed Talk
Abstract : When solving Hamiltonian systems using numerical integrators, preserving the symplectic structure is crucial. We analyze whether the same is true if neural networks (NN) are used. In order to include the symplectic structure in the NN's topology we formulate a generalized framework for two well-known NN topologies and discover a novel topology outperforming all others. We find that symplectic NNs generalize better and give more accurate long-term predictions than physics-unaware NNs.