Message Passing Neural PDE Solvers

Abstract

The numerical solution of partial differential equations (PDEs) is difficult, having led to a century of research so far. Recently, there have been pushes to build neural--numerical hybrid solvers, which piggy-backs the modern trend towards fully end-to-end learned systems. Most works so far can only generalize over a subset of properties to which a generic solver would be faced, including: resolution, topology, geometry, boundary conditions, domain discretization regularity, dimensionality, etc. In this work, we build a solver, satisfying these properties, where all the components are based on neural message passing, replacing all heuristically designed components in the computation graph with backprop-optimized neural function approximators. We show that neural message passing solvers representationally contain some classical methods, such as finite differences, finite volumes, and WENO schemes. In order to encourage stability in training autoregressive models, we put forward a method that is based on the principle of zero-stability, posing stability as a domain adaptation problem. We validate our method on various fluid-like flow problems, demonstrating fast, stable, and accurate performance across different domain topologies, equation parameters, discretizations, etc., in 1D and 2D.

Published at ICLR 2022 (Spotlight paper), Github: https://github.com/brandstetter-johannes/MP-Neural-PDE-Solvers

Cited by in corpus (13)

• Graph Neural Networks Accelerated Molecular Dynamics
• Plasma Surrogate Modelling using Fourier Neural Operators
• Fast Dynamic 1D Simulation of Divertor Plasmas with Neural PDE Surrogates
• Spectrally Adapted Physics-Informed Neural Networks for Solving Unbounded Domain Problems
• Operator Learning with Neural Fields: Tackling PDEs on General Geometries
• Continuous PDE Dynamics Forecasting with Implicit Neural Representations
• LordNet: An Efficient Neural Network for Learning to Solve Parametric Partial Differential Equations without Simulated Data
• Deciphering and integrating invariants for neural operator learning with various physical mechanisms
• Predicting Change, Not States: An Alternate Framework for Neural PDE Surrogates
• Learning the dynamics of a one-dimensional plasma model with graph neural networks
• Variational Monte Carlo Approach to Partial Differential Equations with Neural Networks
• Stability of implicit neural networks for long-term forecasting in dynamical systems
• High Throughput Training of Deep Surrogates from Large Ensemble Runs