Physics-informed neural networks (PINNs): an introductory crash-course

By Ben Moseley, 2022

This workshop builds upon my blog post on PINNs: https://benmoseley.blog/my-research/so-what-is-a-physics-informed-neural-network/.

Read the seminal PINN papers here and here.

Workshop goals

By the end of this workshop, you should be able to:

• code a PINN from scratch in PyTorch
• understand the different types of scientific tasks PINNs can be used for
• understand in more detail how PINNs are trained and how to improve their convergence

Task overview

We will be coding a PINN from scratch in PyTorch and using it solve simulation and inversion tasks related to the damped harmonic oscillator.