This repository contains source codes prepared for Master Thesis done at the AGH UST in Cracow. The code is under refactoring process now.
Topic of the Thesis: Neural networks in solving differential equations
Text of the Thesis is available here
Two approaches were used:
- Physics Informed deep Neural Networks (PINNs) for solving Schrödinger equations for model problems in quantum chemistry, used library: DeepXDE
- Acceleration of DFTB based molecular dynamics (MD) simulation by prediction of atomic charges by SchNet, used library: Schnetpack