Hybrid Finite Difference with the Physics-informed Neural Network for solving PDE in complex geometries

Abstract

The physics-informed neural network (PINN) is effective in solving the partial differential equation (PDE) by capturing the physics constraints as a part of the training loss function through the Automatic Differentiation (AD). This study proposes the hybrid finite difference with the physics-informed neural network (HFD-PINN) to fully use the domain knowledge. The main idea is to use the finite difference method (FDM) locally instead of AD in the framework of PINN. In particular, we use AD at complex boundaries and the FDM in other domains. The hybrid learning model shows promising results in experiments. To use the FDM locally in the complex boundary domain and avoid the generation of background mesh, we propose the HFD-PINN-sdf method, which locally uses the finite difference scheme at random points. In addition, the signed distance function is used to avoid the difference scheme from crossing the domain boundary. In this paper, we demonstrate the performance of our proposed methods and compare the results with the different number of collocation points for the Poisson equation, Burgers equation. We also chose several different finite difference schemes, including the compact finite difference method (CDM) and crank-nicolson method (CNM), to verify the robustness of HFD-PINN. We take the heat conduction problem and the heat transfer problem on the irregular domain as examples to demonstrate the efficacy of our framework. In summary, HFD-PINN, especially HFD-PINN-sdf, are more instructive and efficient, significantly when solving PDEs in complex geometries.

Physics-informed neural network for turbulent flow reconstruction in composite porous-fluid systems

“…In addition, it is necessary to enhance the PINN model by modifying the method for calculating the velocity gradients. This could include hybrid differentiation techniques, combining finite difference or volume method with AD, and applying near-wall velocity gradient treatments similar to conventional numerical methods [44,71,72]. These improvements would enhance the accuracy of both mean flow variables and their gradients.…”

Stiff-PDEs and Physics-Informed Neural Networks

“…NVIDIA Modulus and several other literature takes this one step further by using signed distance function (SDF) weights [74,78,79]. SDF weights are used to assign minuscule weights around the region with conflicting BCs.…”

Numerical modeling of the propagation process of landslide surge using physics-informed deep learning

“…Considering the different mechanical properties of blocks and fluids, some scholars have coupled the Computational Fluid Dynamics (CFD) and discontinuous numerical methods [17], such as the coupled CFD-DEM Approach [18] and the hybrid DEM-SPH Model [19], etc. When solving partial differential equations (PDE), numerical methods such as the Finite Difference Method (FDM) [20] and the Finite Element Method (FEM) [21] are mostly adopted. Most of these numerical methods need to be discretized when solving complex PDEs.…”