Integrated circuits (ICs) rely on materials, devices, and systems governed by complex physical principles, often described by ordinary or partial differential equations (ODEs/PDEs). However, solving these equations becomes increasingly challenging due to growing system complexity, multi-scale dynamics, and limited experimental data for parameter estimation. Physics-informed neural networks (PINNs) have emerged as a promising approach to overcome these challenges by integrating physical laws into machine learning models.

In this project, the student will explore the fundamentals of system dynamics (e.g., spintronic systems modeled by the Landau-Lifshitz-Gilbert (LLG) equations) and investigate how PINNs can be applied to efficiently simulate and analyze such systems. The project will involve numerical modeling, neural network development, and performance evaluation to assess accuracy, scalability, and computational efficiency.