Learning data driven discretizations for partial differential equations

Abstract

The numerical solution of partial differential equations (PDEs) is challenging because of the need to resolve spatiotemporal features over wide length and timescales. Often, it is computationally intractable to resolve the finest features in the solution. The only recourse is to use approximate coarse-grained representations, which aim to accurately represent long-wavelength dynamics while properly accounting for unresolved small scale physics. Deriving such coarse grained equations is notoriously difficult, and often \emph{ad hoc}. Here we introduce \emph{data driven discretization}, a method for learning optimized approximations to PDEs based on actual solutions to the known underlying equations. Our approach uses neural networks to estimate spatial derivatives, which are optimized end-to-end to best satisfy the equations on a low resolution grid. The resulting numerical methods are remarkably accurate, allowing us to integrate in time a collection of nonlinear equations in one spatial dimension at resolutions 4-8x coarser than is possible with standard finite difference methods.

YBS and SH contributed equally to this work. 7 pages, 4 figures (+ Appendix: 9 pages, 10 figures)

References in corpus (12)

• Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
• Axiomatic Attribution for Deep Networks
• Learning Important Features Through Propagating Activation Differences
• PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
• Deep learning for universal linear embeddings of nonlinear dynamics
• Direct numerical simulation of turbulent channel flow up to $R{e}_{\tau }\approx 5200$
• Deep learning to represent sub-grid processes in climate models
• Deep Neural Networks for Data-Driven Turbulence Models
• Accelerating Eulerian Fluid Simulation With Convolutional Networks
• Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge
• Deep Fluids: A Generative Network for Parameterized Fluid Simulations
• Deep Dynamical Modeling and Control of Unsteady Fluid Flows

Cited by in corpus (122)

• Machine learning accelerated computational fluid dynamics
• Data-driven discovery of coordinates and governing equations
• Enhancing Computational Fluid Dynamics with Machine Learning
• WeatherBench: A benchmark dataset for data-driven weather forecasting
• Enforcing Analytic Constraints in Neural-Networks Emulating Physical Systems
• Machine Learning in Aerodynamic Shape Optimization
• Stiff-PINN: Physics-Informed Neural Network for Stiff Chemical Kinetics
• PhyCRNet: Physics-informed Convolutional-Recurrent Network for Solving Spatiotemporal PDEs
• Kohn-Sham equations as regularizer: building prior knowledge into machine-learned physics
• Deep Learning in Deterministic Computational Mechanics
• Data-driven rogue waves and parameter discovery in the defocusing NLS equation with a potential using the PINN deep learning
• A nonlocal physics-informed deep learning framework using the peridynamic differential operator
• Deep learning for dynamic modeling and coded information storage of vector-soliton pulsations in mode-locked fiber lasers
• The Random Feature Model for Input-Output Maps between Banach Spaces
• Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solvers
• JAX-FLUIDS: A fully-differentiable high-order computational fluid dynamics solver for compressible two-phase flows
• Hamiltonian neural networks for solving equations of motion
• A Gradient-based Deep Neural Network Model for Simulating Multiphase Flow in Porous Media
• Stable a posteriori LES of 2D turbulence using convolutional neural networks: Backscattering analysis and generalization to higher Re via transfer learning
• A physics-informed operator regression framework for extracting data-driven continuum models
• Weak baselines and reporting biases lead to overoptimism in machine learning for fluid-related partial differential equations
• Physics-constrained, low-dimensional models for MHD: First-principles and data-driven approaches
• Cellular automata as convolutional neural networks
• Deep Learning for One-dimensional Consolidation
• Predicting the dynamic process and model parameters of the vector optical solitons in birefringent fibers via the modified PINN
• A deep learning framework for solution and discovery in solid mechanics
• Learning hydrodynamic equations for active matter from particle simulations and experiments
• Coarse-scale PDEs from fine-scale observations via machine learning
• Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics
• Learned discretizations for passive scalar advection in a 2-D turbulent flow
• Data-driven peakon and periodic peakon travelling wave solutions of some nonlinear dispersive equations via deep learning
• Enhancement of shock-capturing methods via machine learning
• Optimizing a DIscrete Loss (ODIL) to solve forward and inverse problems for partial differential equations using machine learning tools
• Revealing the state space of turbulence using machine learning
• Neural Galerkin Schemes with Active Learning for High-Dimensional Evolution Equations
• Learning to Control PDEs with Differentiable Physics
• Hyperbolic compactification of M-theory and de Sitter quantum gravity
• Linking Machine Learning with Multiscale Numerics: Data-Driven Discovery of Homogenized Equations
• Machine Learning of Implicit Combinatorial Rules in Mechanical Metamaterials
• A finite element-based physics-informed operator learning framework for spatiotemporal partial differential equations on arbitrary domains
• Deep reconstruction of strange attractors from time series
• Differentiable Molecular Simulations for Control and Learning
• FiniteNet: A Fully Convolutional LSTM Network Architecture for Time-Dependent Partial Differential Equations
• Deep learning probability flows and entropy production rates in active matter
• Data-driven discoveries of Bäcklund transforms and soliton evolution equations via deep neural network learning schemes
• Accelerating high order discontinuous Galerkin solvers using neural networks: 3D compressible Navier-Stokes equations
• Solving and Learning Nonlinear PDEs with Gaussian Processes
• Large Scale Distributed Linear Algebra With Tensor Processing Units
• Frame invariant neural network closures for Kraichnan turbulence
• ML for Flood Forecasting at Scale
• Operator Learning Using Random Features: A Tool for Scientific Computing
• PDE-GCN: Novel Architectures for Graph Neural Networks Motivated by Partial Differential Equations
• Unsupervised Discovery of Inertial-Fusion Plasma Physics using Differentiable Kinetic Simulations and a Maximum Entropy Loss Function
• Low-rank statistical finite elements for scalable model-data synthesis
• Physics-Coupled Neural Network Magnetic Resonance Electrical Property Tomography (MREPT) for Conductivity Reconstruction
• Machine learning based data-driven discovery of nonlinear phase-field dynamics
• Deep Learning of Conjugate Mappings
• towards a robust detection of viscous and turbulent flow regions using unsupervised machine learning
• Surface Similarity Parameter: A New Machine Learning Loss Metric for Oscillatory Spatio-Temporal Data
• Physics-informed attention-based neural network for solving non-linear partial differential equations
• Personalized Algorithm Generation: A Case Study in Learning ODE Integrators
• LordNet: An Efficient Neural Network for Learning to Solve Parametric Partial Differential Equations without Simulated Data
• Explicit physics-informed neural networks for non-linear upscaling closure: the case of transport in tissues
• On the locality of local neural operator in learning fluid dynamics
• GoRINNs: Godunov-Riemann Informed Neural Networks for Learning Hyperbolic Conservation Laws
• Neural Control of Discrete Weak Formulations: Galerkin, Least-Squares and Minimal-Residual Methods with Quasi-Optimal Weights
• Functional Tensor Network Solving Many-body Schrödinger Equation
• Local neural operator for solving transient partial differential equations on varied domains
• Learning normal form autoencoders for data-driven discovery of universal,parameter-dependent governing equations
• LatticeGraphNet: A two-scale graph neural operator for simulating lattice structures
• Machine learning depinning of dislocation pileups
• Distributed computing for physics-based data-driven reduced modeling at scale: Application to a rotating detonation rocket engine
• Machine Learning Changes the Rules for Flux Limiters
• Approximately well-balanced Discontinuous Galerkin methods using bases enriched with Physics-Informed Neural Networks
• Structure Preservation for the Deep Neural Network Multigrid Solver
• Error-correcting neural networks for semi-Lagrangian advection in the level-set method
• Computational model discovery with reinforcement learning
• Physics-guided hierarchical neural networks for Maxwell's equations in plasmonic metamaterials
• Multigrid-Augmented Deep Learning Preconditioners for the Helmholtz Equation using Compact Implicit Layers
• Reinforcement Learning for Adaptive Mesh Refinement
• Variational Data Assimilation with a Learned Inverse Observation Operator
• Godunov Loss Functions for Modelling of Hyperbolic Conservation Laws
• Promoting global stability in data-driven models of quadratic nonlinear dynamics
• Predicting Change, Not States: An Alternate Framework for Neural PDE Surrogates
• Inundation Modeling in Data Scarce Regions
• Frequency-compensated PINNs for Fluid-dynamic Design Problems
• Feature Engineering with Regularity Structures
• Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning
• Particles to Partial Differential Equations Parsimoniously
• Least-Squares Neural Network (LSNN) Method For Linear Advection-Reaction Equation: Discontinuity Interface
• A learned conservative semi-Lagrangian finite volume scheme for transport simulations
• Rapidly Encoding Generalizable Dynamics in a Euclidean Symmetric Neural Network
• Reducing Spatial Discretization Error on Coarse CFD Simulations Using an OpenFOAM-Embedded Deep Learning Framework
• Unsupervised Reservoir Computing for Solving Ordinary Differential Equations
• A Deep Learning Approach for Predicting Spatiotemporal Dynamics From Sparsely Observed Data
• Physics-Guided Discovery of Highly Nonlinear Parametric Partial Differential Equations
• A composable autoencoder-based iterative algorithm for accelerating numerical simulations
• A reduced order Schwarz method for nonlinear multiscale elliptic equations based on two-layer neural networks
• Learnable-Differentiable Finite Volume Solver for Accelerated Simulation of Flows
• Engineering morphogenesis of cell clusters with differentiable programming
• Physics-aware Spatiotemporal Modules with Auxiliary Tasks for Meta-Learning
• STENCIL-NET: Data-driven solution-adaptive discretization of partial differential equations
• Numerical analysis of a deep learning formulation of elastic full waveform inversion with high order total variation regularization in different parameterization
• Learning Runge-Kutta Integration Schemes for ODE Simulation and Identification
• The Seven-League Scheme: Deep learning for large time step Monte Carlo simulations of stochastic differential equations
• All-optical photoacoustic tomography via beam deflection
• Machine learning and serving of discrete field theories -- when artificial intelligence meets the discrete universe
• Self-assembling kinetics: Accessing a new design space via differentiable statistical-physics models
• Tuning Colloidal Reactions
• An asynchronous discontinuous Galerkin method for massively parallel PDE solvers
• Discovering Hidden Physics Behind Transport Dynamics
• ISALT: Inference-based schemes adaptive to large time-stepping for locally Lipschitz ergodic systems
• A Latent space solver for PDE generalization
• Predicting the Dynamics of Complex System via Multiscale Diffusion Autoencoder
• Neural network-based Godunov corrections for approximate Riemann solvers using bi-fidelity learning
• Machine Learning for Discovering Effective Interaction Kernels between Celestial Bodies from Ephemerides
• SlotPi: Physics-informed Object-centric Reasoning Models
• Physics-Aware Downsampling with Deep Learning for Scalable Flood Modeling
• Object classification in analytical chemistry via data-driven discovery of partial differential equations
• Towards Exascale Design of Soft Mesoscale Materials
• NySALT: Nyström-type inference-based schemes adaptive to large time-stepping
• Learning Compact Physics-Aware Delayed Photocurrent Models Using Dynamic Mode Decomposition