Physics-informed reduced order model with conditional neural fields
Preprint
Abstract
This study presents the conditional neural fields for reduced-order modeling (CNF-ROM) framework to approximate solutions of parametrized partial differential equations (PDEs). The approach combines a parametric neural ODE (PNODE) for modeling latent dynamics over time with a decoder that reconstructs PDE solutions from the corresponding latent states. We introduce a physics-informed learning objective for CNF-ROM, which includes two key components. First, the framework uses coordinate-based neural networks to calculate and minimize PDE residuals by computing spatial derivatives via automatic differentiation and applying the chain rule for time derivatives. Second, exact initial and boundary conditions (IC/BC) are imposed using approximate distance functions (ADFs) [Sukumar and Srivastava, CMAME, 2022]. However, ADFs introduce a trade-off as their second- or higher-order derivatives become unstable at the joining points of boundaries. To address this, we introduce an auxiliary network inspired by [Gladstone et al., NeurIPS ML4PS workshop, 2022]. Our method is validated through parameter extrapolation and interpolation, temporal extrapolation, and comparisons with analytical solutions.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- Enforcing Dirichlet boundary conditions in physics-informed neural networks and variational physics-informed neural networksHeliyon, 2023
- Neural Fields in Visual Computing and BeyondComputer Graphics Forum, 2022
- Exact imposition of boundary conditions with distance functions in physics-informed deep neural networksComputer Methods in Applied Mechanics and Engineering, 2022
- A fast and accurate physics-informed neural network reduced order model with shallow masked autoencoderJournal of Computational Physics, 2022
- Reduced order models for Lagrangian hydrodynamicsComputer Methods in Applied Mechanics and Engineering, 2022
- Physics-informed neural networks (PINNs) for fluid mechanics: a reviewActa Mechanica Sinica, 2021
- Parameterized neural ordinary differential equations: applications to computational physics problemsProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2021
- Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencodersJournal of Computational Physics, 2020
- Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equationsJournal of Computational Physics, 2019
- The GNAT method for nonlinear model reduction: Effective implementation and application to computational fluid dynamics and turbulent flowsJournal of Computational Physics, 2013