Study on a 3D Possion's Equation Slover Based on Deep Learning Technique

Abstract

In this study, we investigate the feasibility of applying deep learning technique to build a 3D electrostatic solver. A deep convolutional neural network (CNN) is proposed to take advantage of the power of CNN in approximation of highly nonlinear functions and prediction of the potential distribution of electrostatic field. Compared with traditional numerical solvers based on finite difference scheme, this method uses a data-driven end-to-end model. Numerical experiments show that the prediction error can reach below 3 percent and the computing time can be significantly reduced compared with traditional finite difference solvers.