Spectrally Adapted Physics-Informed Neural Networks for Solving Unbounded Domain Problems
arXiv:2202.02710 · doi:10.1088/2632-2153/acd0a1
Abstract
Solving analytically intractable partial differential equations (PDEs) that involve at least one variable defined on an unbounded domain arises in numerous physical applications. Accurately solving unbounded domain PDEs requires efficient numerical methods that can resolve the dependence of the PDE on the unbounded variable over at least several orders of magnitude. We propose a solution to such problems by combining two classes of numerical methods: (i) adaptive spectral methods and (ii) physics-informed neural networks (PINNs). The numerical approach that we develop takes advantage of the ability of physics-informed neural networks to easily implement high-order numerical schemes to efficiently solve PDEs and extrapolate numerical solutions at any point in space and time. We then show how recently introduced adaptive techniques for spectral methods can be integrated into PINN-based PDE solvers to obtain numerical solutions of unbounded domain problems that cannot be efficiently approximated by standard PINNs. Through a number of examples, we demonstrate the advantages of the proposed spectrally adapted PINNs in solving PDEs and estimating model parameters from noisy observations in unbounded domains.
arXiv comment29 pages, 8 figures
References in corpus (23)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators
- B-PINNs: Bayesian Physics-Informed Neural Networks for Forward and Inverse PDE Problems with Noisy Data
- Fourier Neural Operator for Parametric Partial Differential Equations
- Dedalus: A Flexible Framework for Numerical Simulations with Spectral Methods
- Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations
- PDE-Net: Learning PDEs from Data
- Variational Physics-Informed Neural Networks For Solving Partial Differential Equations
- Regularization for Deep Learning: A Taxonomy
- Physics-Informed Neural Operator for Learning Partial Differential Equations
- Symplectic ODE-Net: Learning Hamiltonian Dynamics with Control
- Toward an AI Physicist for Unsupervised Learning
- Implicit energy regularization of neural ordinary-differential-equation control
- Modeling System Dynamics with Physics-Informed Neural Networks Based on Lagrangian Mechanics
- Message Passing Neural PDE Solvers
- Recipes for when Physics Fails: Recovering Robust Learning of Physics Informed Neural Networks
- Algebraic damping in the one-dimensional Vlasov equation
- Neural Ordinary Differential Equation Control of Dynamics on Graphs
- Adversarial Multi-task Learning Enhanced Physics-informed Neural Networks for Solving Partial Differential Equations
- Noise-aware Physics-informed Machine Learning for Robust PDE Discovery
- Near-optimal control of dynamical systems with neural ordinary differential equations
- Kinetic theory for structured populations: application to stochastic sizer-timer models of cell proliferation
- Modeling Deformed Transmission Lines for Continuous Strain Sensing Applications