Physics-Informed Neural Networks with PyTorch

This repository explores Physics-Informed Neural Networks (PINNs) using PyTorch.
PINNs integrate physical laws, expressed as partial differential equations (PDEs), into the training of neural networks.
This approach is especially useful when data is scarce but the governing physics is well-understood.

Overview

This project demonstrates how to implement PINNs to solve PDEs by embedding physical constraints directly into the loss function of a neural network.
The learned solutions not only fit the data but also respect the underlying physical laws.

📝 For a detailed tutorial, check out the accompanying Medium article.

Repository Structure

pinns/
├── src/              # Core implementation: models, training loops, etc.
├── LICENSE
└── README.md         # You're here!

Installation

Requirements

Install the required packages using pip:

pip install torch scikit-learn numpy matplotlib seaborn

Clone the Repository

git clone https://github.com/TheodoreWolf/pinns.git
cd pinns

Usage

To get started, open one of the Jupyter notebooks in the src/ directory:

jupyter notebook src/

These walk through training PINNs on various PDEs with visualizations.

Alternatively, run Python scripts in src/ to train directly via the command line.

Resources

• 📘 Medium Tutorial: Physics-Informed Neural Networks: A Simple Tutorial with PyTorch
• 📄 Original Paper: Raissi, Perdikaris, & Karniadakis (2019)
  Physics-informed neural networks (PINNs)

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

Inspired by the foundational work by Raissi et al.
This repo aims to provide an approachable and practical introduction to PINNs with PyTorch.