Repository for the code associated with the guest lecture presented at the University of Missouri "Introduction to Physics Informed Neural Networks (PINNS)", Atomistic Materials Analytics, EECS 8615.
Installation of the required packages to run the provided code will vary based on your system.
I prefer to use Conda to manage my python environments. However, if you
use something different, refer to the requirements.txt file for the required packages.
NOTE: As of 05/08/25 the most recent version of PyTorch is not compatible with numpy>=2.
conda create -n pinn python=3.12
conda activate pinn
python -m pip install -r requirements.txtIt's not necessarily good practice to use pip with conda, however it seems to be
the best option for this project.
Please reach out if there are issues with installation and I will provide a dockerfile and docker image to run the code in a container. Might do this anyway.
I can't recommend Dr. Ben Moseley's blog/github/published work enough when it comes to learning about PINNs. He is my first go to when I need to refresh my memory on the subject. With that said, this code is based on the following references:
https://beltoforion.de/en/harmonic_oscillator/
Lin, Xing, Yair Rivenson, Nezih T. Yardimci, Muhammed Veli, Yi Luo, Mona Jarrahi, and Aydogan Ozcan. 2018. “All-Optical Machine Learning Using Diffractive Deep Neural Networks.” Science 361 (6406): 1004–8. https://doi.org/10.1126/science.aat8084.
G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang, “Physics-informed machine learning,” Nat Rev Phys, vol. 3, no. 6, pp. 422–440, May 2021, doi: 10.1038/s42254-021-00314-5.
Chen, Z., Liu, Y. & Sun, H. Physics-informed learning of governing equations from scarce data. Nat Commun 12, 6136 (2021). https://doi.org/10.1038/s41467-021-26434-1
Raissi et al, Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear and partial differential equations, JCP (2018)