Learning to Simulate Complex Physics with Graph Networks

TL;DR

This paper introduces Graph Network-based Simulators (GNS), a machine learning framework that learns to simulate complex physical systems involving fluids, solids, and deformable materials, demonstrating strong generalization and robustness.

Contribution

The paper presents a novel GNS framework that effectively learns to simulate diverse physical phenomena with generalization to larger scales and different initial conditions, advancing physical simulation methods.

Findings

Model generalizes from single-timestep training to long-term predictions

Robust to hyperparameter variations and noise in training data

Achieves state-of-the-art performance in learned physical simulation

Abstract

Here we present a machine learning framework and model implementation that can learn to simulate a wide variety of challenging physical domains, involving fluids, rigid solids, and deformable materials interacting with one another. Our framework---which we term "Graph Network-based Simulators" (GNS)---represents the state of a physical system with particles, expressed as nodes in a graph, and computes dynamics via learned message-passing. Our results show that our model can generalize from single-timestep predictions with thousands of particles during training, to different initial conditions, thousands of timesteps, and at least an order of magnitude more particles at test time. Our model was robust to hyperparameter choices across various evaluation metrics: the main determinants of long-term performance were the number of message-passing steps, and mitigating the accumulation of error by corrupting the training data with noise. Our GNS framework advances the state-of-the-art in learned physical simulation, and holds promise for solving a wide range of complex forward and inverse problems.