Solving Elliptic Problems with Singular Sources using Singularity Splitting Deep Ritz Method

09/07/2022
βˆ™
by   Tianhao Hu, et al.
βˆ™
0
βˆ™

In this work, we develop an efficient solver based on deep neural networks for the Poisson equation with variable coefficients and singular sources expressed by the Dirac delta function δ(𝐱). This class of problems covers general point sources, line sources and point-line combinations, and has a broad range of practical applications. The proposed approach is based on decomposing the true solution into a singular part that is known analytically using the fundamental solution of the Laplace equation and a regular part that satisfies a suitable elliptic PDE with smoother sources, and then solving for the regular part using the deep Ritz method. A path-following strategy is suggested to select the penalty parameter for penalizing the Dirichlet boundary condition. Extensive numerical experiments in two- and multi-dimensional spaces with point sources, line sources or their combinations are presented to illustrate the efficiency of the proposed approach, and a comparative study with several existing approaches is also given, which shows clearly its competitiveness for the specific class of problems. In addition, we briefly discuss the error analysis of the approach.

READ FULL TEXT

page 12

page 13

page 15

page 17

page 18

research
βˆ™ 12/22/2020

Regularity and finite element approximation for two-dimensional elliptic equations with line Dirac sources

We study the elliptic equation with a line Dirac delta function as the s...
research
βˆ™ 10/01/2019

A regularization approach for solving Poisson's equation with singular charge sources and diffuse interfaces

Singular charge sources in terms of Dirac delta functions present a well...
research
βˆ™ 08/24/2023

Dirichlet-Neumann and Neumann-Neumann Methods for Elliptic Control Problems

We present the Dirichlet-Neumann (DN) and Neumann-Neumann (NN) methods a...
research
βˆ™ 08/31/2023

Solving Poisson Problems in Polygonal Domains with Singularity Enriched Physics Informed Neural Networks

Physics informed neural networks (PINNs) represent a very powerful class...
research
βˆ™ 10/11/2022

An efficient neural-network and finite-difference hybrid method for elliptic interface problems with applications

A new and efficient neural-network and finite-difference hybrid method i...
research
βˆ™ 03/29/2023

Conductivity Imaging from Internal Measurements with Mixed Least-Squares Deep Neural Networks

In this work we develop a novel approach using deep neural networks to r...
research
βˆ™ 06/13/2022

Box constraints and weighted sparsity regularization for identifying sources in elliptic PDEs

We explore the possibility for using boundary data to identify sources i...

Please sign up or login with your details

Forgot password? Click here to reset