We propose a deep neural network architecture designed such that its output forms an invertible symplectomorphism of the input. This design draws an analogy to the real-valued non-volume-preserving (real NVP) method used in normalizing flow techniques. Utilizing this neural network type allows for learning tasks on unknown Hamiltonian systems without breaking the inherent symplectic structure of the phase space.
Preservation of stability properties near fixed points of linear hamiltonian systems by symplectic integrators
Harbin Institute of Technology(哈尔滨工业大学);University of Washington(华盛顿大学);Academy of Mathematics and System Sciences, Chinese Academy of Sciences(中国科学院数学与系统科学研究院)
Preservation of Physical Properties of Stochastic Maxwell Equations with Additive Noise via Stochastic Multi-symplectic Methods
Institute of Computational Mathematics and Scientific/Engineering Computing, Academy of Mathematics and Systems Science, Chinese Academy of Sciences(中国科学院数学与系统科学研究院计算数学与科学工程计算研究所)
Reduced-order modeling of Hamiltonian dynamics based on symplectic neural networks
School of Mathematical Sciences, Zhejiang University(浙江大学数学科学学院);Department of Mathematics and Statistics, Texas Tech University(德克萨斯理工大学数学与统计系);School of Computational Science and Engineering, Georgia Institute of Technology(佐治亚理工学院计算科学与工程学院)