Physics-informed neural networks (PINNs)

AI
Author

Yiluan Song

Published

July 27, 2025

I prepared a series of teaching materials on physics-informed neural networks (PINNs) for U-M MIDAS AI for Scientists and Engineers Summer Academy 2025 and pre-workshop materials for U-M MIDAS-KGML workshop.

Slides

A high-level overview of PINN concepts and motivation. Start here to understand the big picture and foundational ideas before diving into the code.

Part of the slides were developed based on slides by Dr. Alexander Rodríguez.

Notebooks

Explore hands-on PINN tutorials: - Notebook 1: Harmonic oscillator (a simple example) - Notebook 2: Cyclic voltammetry (a forward problem) - Notebook 3: Fluid dynamics (an inverse problem)

Each notebook is self-contained and designed to run in Google Colab or Jupyter.

The tutorials were adapted from tutorials by Dr. Ben Moseley, Dr. Haotian Chen, and Dr. Seungchul Lee.

Recordings

Part 1

Part 2