Physics-Informed Neural Networks with good lattice points set for solving low regularity PDEs and high dimensional PDE

When dealing with a large number of points was required, the traditional uniform sampling approach for approximating integrals using the Monte Carlo method becomes inefficient. In this work, we leverage the good lattice point sets from number-theoretic methods for sampling purposes and develop a deep learning framework that integrates the good lattice point sets with Physics-Informed Neural Networks. This framework is designed to address low-regularity and high-dimensional problems. Furthermore, rigorous mathematical proofs are provided for our algorithm, demonstrating its validity. Lastly, in the experimental section, we employ numerical experiments involving the Poisson equation with low regularity, the two-dimensional inverse Helmholtz equation, and high-dimensional linear and nonlinear problems to illustrate the effectiveness of our algorithm from a numerical perspective.