ABSTRACT
Computational physics is a new subject that uses computers to numerically simulate physical processes. The application of computational physics is usually fairly extensive and permeates all fields of physics. The research process of computational physics mainly includes modeling, simulation, and computing. Using the method of moments to solve the integral equation needs to be subject to certain conditions. Deep learning technology has created many brilliant achievements in the field of computational physics. This chapter discusses several basic architectures in depth. In fact, there are many paradigms to solve physical problems using machine learning. This part mainly discusses four of them: data-driven, physical constraints, operator learning and deep learning-traditional algorithm fusion. In the field of partial differential equations, traditional numerical methods utilize discrete structures to approximate the mapping of infinite dimensional operators to finite dimensional approximation problems. The chapter also presents an overview on the key concepts discussed in this book.
