Finite element interpolated neural networks for solving forward and inverse problems
From MaRDI portal
Publication:6118570
DOI10.1016/J.CMA.2023.116505arXiv2306.06304OpenAlexW4387443591MaRDI QIDQ6118570FDOQ6118570
Authors: Santiago Badia, Wei Li, Alberto F. Martín 
Publication date: 21 March 2024
Published in: Computer Methods in Applied Mechanics and Engineering (Search for Journal in Brave)
Abstract: We propose a general framework for solving forward and inverse problems constrained by partial differential equations, where we interpolate neural networks onto finite element spaces to represent the (partial) unknowns. The framework overcomes the challenges related to the imposition of boundary conditions, the choice of collocation points in physics-informed neural networks, and the integration of variational physics-informed neural networks. A numerical experiment set confirms the framework's capability of handling various forward and inverse problems. In particular, the trained neural network generalises well for smooth problems, beating finite element solutions by some orders of magnitude. We finally propose an effective one-loop solver with an initial data fitting step (to obtain a cheap initialisation) to solve inverse problems.
Full work available at URL: https://arxiv.org/abs/2306.06304
Cites Work
- Gmsh: a 3-D finite element mesh generator with built-in pre- and post-processing facilities
- The software design of gridap: a finite element package based on the Julia JIT compiler
- Title not available (Why is that?)
- Optimization with PDE Constraints
- Finite element exterior calculus, homological techniques, and applications
- A posteriori error estimation in finite element analysis
- Preconditioning discretizations of systems of partial differential equations.
- A second-order-accurate symmetric discretization of the Poisson equation on irregular domains
- On quadrature rules for solving partial differential equations using neural networks
- Hybrid FEM-NN models: combining artificial neural networks with the finite element method
- Composing scalable nonlinear algebraic solvers
- Multilevel balancing domain decomposition at extreme scales
- The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
- Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- A penalty method for PDE-constrained optimization in inverse problems
- Extended Physics-Informed Neural Networks (XPINNs): A Generalized Space-Time Domain Decomposition Based Deep Learning Framework for Nonlinear Partial Differential Equations
- Scheduling massively parallel multigrid for multilevel Monte Carlo methods
- Exact imposition of boundary conditions with distance functions in physics-informed deep neural networks
- A tutorial on the adjoint method for inverse problems
- A comparison study of deep Galerkin method and deep Ritz method for elliptic problems with different boundary conditions
- Variational physics informed neural networks: the role of quadratures and test functions
- Finite basis physics-informed neural networks (FBPINNs): a scalable domain decomposition approach for solving differential equations
Cited In (2)
This page was built for publication: Finite element interpolated neural networks for solving forward and inverse problems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6118570)
