Efficient physics-informed neural networks using hash encoding
From MaRDI portal
Abstract: Physics-informed neural networks (PINNs) have attracted a lot of attention in scientific computing as their functional representation of partial differential equation (PDE) solutions offers flexibility and accuracy features. However, their training cost has limited their practical use as a real alternative to classic numerical methods. Thus, we propose to incorporate multi-resolution hash encoding into PINNs to improve the training efficiency, as such encoding offers a locally-aware (at multi resolution) coordinate inputs to the neural network. Borrowed from the neural representation field community (NeRF), we investigate the robustness of calculating the derivatives of such hash encoded neural networks with respect to the input coordinates, which is often needed by the PINN loss terms. We propose to replace the automatic differentiation with finite-difference calculations of the derivatives to address the discontinuous nature of such derivatives. We also share the appropriate ranges for the hash encoding hyperparameters to obtain robust derivatives. We test the proposed method on three problems, including Burgers equation, Helmholtz equation, and Navier-Stokes equation. The proposed method admits about a 10-fold improvement in efficiency over the vanilla PINN implementation.
Recommendations
- A dimension-augmented physics-informed neural network (DaPINN) with high level accuracy and efficiency
- Diminishing spectral bias in physics-informed neural networks using spatially-adaptive Fourier feature encoding
- Sobolev training for physics-informed neural networks
- Improved training of physics-informed neural networks for parabolic differential equations with sharply perturbed initial conditions
- Residual-based attention in physics-informed neural networks
Cites work
- A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks
- Adaptive fourth-order phase field analysis for brittle fracture
- Bayesian deep convolutional encoder-decoder networks for surrogate modeling and uncertainty quantification
- Exact fully 3D Navier–Stokes solutions for benchmarking
- scientific article; zbMATH DE number 1113267 (Why is no real title available?)
- scientific article; zbMATH DE number 6982909 (Why is no real title available?)
- Multilayer feedforward networks are universal approximators
- On the eigenvector bias of Fourier feature networks: from regression to solving multi-scale PDEs with physics-informed neural networks
- Pattern recognition.
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Scientific machine learning through physics-informed neural networks: where we are and what's next
- Self-adaptive physics-informed neural networks
- Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks
Cited in
(7)- Solving large-scale variational inequalities with dynamically adjusting initial condition in physics-informed neural networks
- Parameterized analytical solution of seepage equation for reservoir simulation using physics-informed Kolmogorov-Arnold network without labels
- Learnable activation functions in physics-informed neural networks for solving partial differential equations
- Weak neural Galerkin method for nonlinear time-dependent partial differential equations
- FE-PIRBN: feature-enhanced physics-informed radial basis neural networks for solving high-frequency electromagnetic scattering problems
- PirateNets: physics-informed deep learning with residual adaptive networks
- Energy loss informed cell-based multilayer perceptron for solving elliptic PDEs with multiscale features
This page was built for publication: Efficient physics-informed neural networks using hash encoding
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6126546)
