Hamiltonian neural networks for solving equations of motion

Abstract

There has been a wave of interest in applying machine learning to study dynamical systems. We present a Hamiltonian neural network that solves the differential equations that govern dynamical systems. This is an equation-driven machine learning method where the optimization process of the network depends solely on the predicted functions without using any ground truth data. The model learns solutions that satisfy, up to an arbitrarily small error, Hamilton's equations and, therefore, conserve the Hamiltonian invariants. The choice of an appropriate activation function drastically improves the predictability of the network. Moreover, an error analysis is derived and states that the numerical errors depend on the overall network performance. The Hamiltonian network is then employed to solve the equations for the nonlinear oscillator and the chaotic Henon-Heiles dynamical system. In both systems, a symplectic Euler integrator requires two orders more evaluation points than the Hamiltonian network in order to achieve the same order of the numerical error in the predicted phase space trajectories.

References in corpus (22)

• Artificial Neural Networks for Solving Ordinary and Partial Differential Equations
• DGM: A deep learning algorithm for solving partial differential equations
• Recent Advances in Physical Reservoir Computing: A Review
• Solving high-dimensional partial differential equations using deep learning
• Turbulence Modeling in the Age of Data
• Deep learning for universal linear embeddings of nonlinear dynamics
• Multivariate LSTM-FCNs for Time Series Classification
• Invariant Variation Problems
• Machine Learning of Linear Differential Equations using Gaussian Processes
• Learning data driven discretizations for partial differential equations
• Data-Driven Forecasting of High-Dimensional Chaotic Systems with Long Short-Term Memory Networks
• Inferring solutions of differential equations using noisy multi-fidelity data
• Deep learning of free boundary and Stefan problems
• On Learning Hamiltonian Systems from Data
• Physics enhanced neural networks predict order and chaos
• Neural Network Models for the Anisotropic Reynolds Stress Tensor in Turbulent Channel Flow
• Modeling the effect of the vaccination campaign on the Covid-19 pandemic
• Port-Hamiltonian Neural Networks for Learning Explicit Time-Dependent Dynamical Systems
• Neural Networks Trained to Solve Differential Equations Learn General Representations
• Adaptable Hamiltonian neural networks
• Learning Hamiltonian dynamics by reservoir computer
• Machine learning with observers predicts complex spatiotemporal behavior

Cited by in corpus (27)

• Physics-informed neural networks for quantum control
• Feasibility Study of Neural ODE and DAE Modules for Power System Dynamic Component Modeling
• Modeling the effect of the vaccination campaign on the Covid-19 pandemic
• Solving the Teukolsky equation with physics-informed neural networks
• Cosmology-informed neural networks to solve the background dynamics of the Universe
• Solving Differential Equations Using Neural Network Solution Bundles
• Using physics-informed neural networks to compute quasinormal modes
• Constrained Hamiltonian Systems and Physics-Informed Neural Networks: Hamilton-Dirac Neural Networks
• Quasinormal Modes in Modified Gravity using Physics-Informed Neural Networks
• Destructive relativity
• Physics-Informed Neural Networks for Nonlocal Beam Eigenvalue Problems
• Optimizing Neural Networks via Koopman Operator Theory
• One-Shot Transfer Learning of Physics-Informed Neural Networks
• Modeling the Time Evolution of Compact Binary Systems with Machine Learning
• Efficient PINNs via Multi-Head Unimodular Regularization of the Solutions Space
• Hamiltonian Neural Networks approach to fuzzball geodesics
• Unsupervised Reservoir Computing for Solving Ordinary Differential Equations
• Joint Parameter Discovery and Generative Modeling of Dynamic Systems
• Embedding stochastic differential equations into neural networks via dual processes
• On tensor invariants of the Clebsch system
• Multi-resolution Physics-Aware Recurrent Convolutional Neural Network for Complex Flows
• Physical Constraint Embedded Neural Networks for inference and noise regulation
• A New Artificial Neuron Proposal with Trainable Simultaneous Local and Global Activation Function
• Error Estimation and Correction from within Neural Network Differential Equation Solvers
• Highly efficient nuclear population transfer through physics-informed neural networks
• Integrability of non-homogeneous Hamiltonian systems with gyroscopic coupling
• Integrated utilization of equations and small dataset in the Koopman operator: applications to forward and inverse problems