Differential equations are widely used in any scientific field. A new method of solving differential equations using deep learning methods has great prospects and have greater versatility over the grid numerical methods. Neural networks can be used as a method for efficiently solving difficult partial differential equations(PDEs). The library that provides the method for solving a general form of PDEs using deep learning will be a useful tool for many scientists and engineers, as well as students.

This project aims to design of a general solver for different types of PDEs using deep learning approach base on the PINNs algorithm as part of NeuralPDE.jl library using the ModelingToolkit.jl PDE interface.

Organization

Student

Kirill Zubov

Mentors

  • Vaibhav Dixit
  • Lu Lu
  • ChrisRackauckas
close

2020