Symplectic Recurrent Neural Networks
arXiv:1909.13334
Abstract
We propose Symplectic Recurrent Neural Networks (SRNNs) as learning algorithms that capture the dynamics of physical systems from observed trajectories. An SRNN models the Hamiltonian function of the system by a neural network and furthermore leverages symplectic integration, multiple-step training and initial state optimization to address the challenging numerical issues associated with Hamiltonian systems. We show that SRNNs succeed reliably on complex and noisy Hamiltonian systems. We also show how to augment the SRNN integration scheme in order to handle stiff dynamical systems such as bouncing billiards.
arXiv commentAdded link to GitHub repository
References in corpus (9)
- Adam: A Method for Stochastic Optimization
- Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
- Interaction Networks for Learning about Objects, Relations and Physics
- Hamiltonian Neural Networks
- A Compositional Object-Based Approach to Learning Physical Dynamics
- Explicit symplectic approximation of nonseparable Hamiltonians: algorithm and long time performance
- AntisymmetricRNN: A Dynamical System View on Recurrent Neural Networks
- Hamiltonian Graph Networks with ODE Integrators
- Recurrent Neural Networks in the Eye of Differential Equations
Cited by in corpus (24)
- A Review of Machine Learning Methods Applied to Structural Dynamics and Vibroacoustic
- Augmenting Physical Models with Deep Networks for Complex Dynamics Forecasting
- Lagrangian Neural Networks
- Thermodynamics-informed graph neural networks
- Physics-Informed Supervised Residual Learning for Electromagnetic Modeling
- Symplectic Neural Networks in Taylor Series Form for Hamiltonian Systems
- A neural network multigrid solver for the Navier-Stokes equations
- Implicit Regularization and Momentum Algorithms in Nonlinearly Parameterized Adaptive Control and Prediction
- Symplectic Learning for Hamiltonian Neural Networks
- Neural Canonical Transformation with Symplectic Flows
- Thermodynamics-informed neural networks for physically realistic mixed reality
- Deep Hamiltonian networks based on symplectic integrators
- Locally-symplectic neural networks for learning volume-preserving dynamics
- Structure Preservation for the Deep Neural Network Multigrid Solver
- Neural Dynamical Systems: Balancing Structure and Flexibility in Physical Prediction
- Sparse Symplectically Integrated Neural Networks
- Machine-learning Kohn-Sham potential from dynamics in time-dependent Kohn-Sham systems
- Forward Prediction for Physical Reasoning
- Discrete Lagrangian Neural Networks with Automatic Symmetry Discovery
- Understanding Recurrent Neural Networks Using Nonequilibrium Response Theory
- MomentumRNN: Integrating Momentum into Recurrent Neural Networks
- MC-LSTM: Mass-Conserving LSTM
- Time-Reversal Symmetric ODE Network
- Neural Symplectic Integrator with Hamiltonian Inductive Bias for the Gravitational -body Problem