Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning
From MaRDI portal
Abstract: Conservation laws are considered to be fundamental laws of nature. It has broad applications in many fields, including physics, chemistry, biology, geology, and engineering. Solving the differential equations associated with conservation laws is a major branch in computational mathematics. The recent success of machine learning, especially deep learning in areas such as computer vision and natural language processing, has attracted a lot of attention from the community of computational mathematics and inspired many intriguing works in combining machine learning with traditional methods. In this paper, we are the first to view numerical PDE solvers as an MDP and to use (deep) RL to learn new solvers. As proof of concept, we focus on 1-dimensional scalar conservation laws. We deploy the machinery of deep reinforcement learning to train a policy network that can decide on how the numerical solutions should be approximated in a sequential and spatial-temporal adaptive manner. We will show that the problem of solving conservation laws can be naturally viewed as a sequential decision-making process, and the numerical schemes learned in such a way can easily enforce long-term accuracy. Furthermore, the learned policy network is carefully designed to determine a good local discrete approximation based on the current state of the solution, which essentially makes the proposed method a meta-learning approach. In other words, the proposed method is capable of learning how to discretize for a given situation mimicking human experts. Finally, we will provide details on how the policy network is trained, how well it performs compared with some state-of-the-art numerical solvers such as WENO schemes, and supervised learning based approach L3D and PINN, and how well it generalizes.
Recommendations
- Conservative physics-informed neural networks on discrete domains for conservation laws: applications to forward and inverse problems
- Learning data-driven discretizations for partial differential equations
- A deep reinforcement learning framework for dynamic optimization of numerical schemes for compressible flow simulations
- Deep reinforcement learning of viscous incompressible flow
- Modeling the dynamics of PDE systems with physics-constrained deep auto-regressive networks
- Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks
- Reinforcement learning-based model reduction for partial differential equations: application to the Burgers equation
- Discretizationnet: a machine-learning based solver for Navier-Stokes equations using finite volume discretization
- Solving non-linear Kolmogorov equations in large dimensions by using deep learning: a numerical comparison of discretization schemes
- Deep learning solver for solving advection-diffusion equation in comparison to finite difference methods
Cites work
- A fast sweeping method for Eikonal equations
- A Multiscale Neural Network Based on Hierarchical Matrices
- A new TVD flux-limiter method for solving nonlinear hyperbolic equations
- An artificial neural network as a troubled-cell indicator
- Constraint-aware neural networks for Riemann problems
- Controlling oscillations in high-order discontinuous Galerkin schemes using artificial viscosity tuned by neural networks
- Deep learning
- DGM: a deep learning algorithm for solving partial differential equations
- Finite Volume Methods for Hyperbolic Problems
- General solutions for nonlinear differential equations: a rule-based self-learning approach using deep reinforcement learning
- scientific article; zbMATH DE number 1302036 (Why is no real title available?)
- scientific article; zbMATH DE number 914383 (Why is no real title available?)
- Learning data-driven discretizations for partial differential equations
- Matching pursuits with time-frequency dictionaries
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Solving high-dimensional partial differential equations using deep learning
- SwitchNet: a neural Network Model for forward and inverse scattering problems
- The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
- Variational training of neural network approximations of solution maps for physical models
- Weighted essentially non-oscillatory schemes
Cited in
(27)- Enhanced fifth order WENO shock-capturing schemes with deep learning
- Lagrangian dual framework for conservative neural network solutions of kinetic equations
- Deep reinforcement learning of viscous incompressible flow
- General solutions for nonlinear differential equations: a rule-based self-learning approach using deep reinforcement learning
- Neural control of discrete weak formulations: Galerkin, least squares \& minimal-residual methods with quasi-optimal weights
- Convergence rate analysis for deep Ritz method
- A deep reinforcement learning framework for dynamic optimization of numerical schemes for compressible flow simulations
- Neural Networks with Local Converging Inputs (NNLCI) for Solving Conservation Laws, Part I: 1D Problems
- Deep reinforcement learning for adaptive mesh refinement
- Learning the nonlinear flux function of a hidden scalar conservation law from data
- Error analysis of deep Ritz methods for elliptic equations
- A Reinforcement Learning Based Slope Limiter for Two-Dimensional Finite Volume Schemes
- Identification of the flux function of nonlinear conservation laws with variable parameters
- A learned conservative semi-Lagrangian finite volume scheme for transport simulations
- Learning the Dynamics for Unknown Hyperbolic Conservation Laws Using Deep Neural Networks
- On mathematical modeling in image reconstruction and beyond
- DynAMO: multi-agent reinforcement learning for dynamic anticipatory mesh optimization with applications to hyperbolic conservation laws
- Data-driven modeling of partially observed biological systems
- Learning based numerical methods for acoustic frequency-domain simulation with high frequency
- Deep learning of first-order nonlinear hyperbolic conservation law solvers
- Cell-average based neural network method for Hunter-Saxton equations
- Machine learning-based WENO5 scheme
- PDE generalization of in-context operator networks: a study on 1D scalar nonlinear conservation laws
- An optimal control deep learning method to design artificial viscosities for discontinuous Galerkin schemes
- Discovering artificial viscosity models for discontinuous Galerkin approximation of conservation laws using physics-informed machine learning
- A third-order finite difference weighted essentially non-oscillatory scheme with shallow neural network
- Deep Ritz methods for Laplace equations with Dirichlet boundary condition
This page was built for publication: Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5162375)
