This repository serves as the official project page for the "AI in the Sciences and Engineering" course at ETH Zurich, offering a comprehensive curriculum and practical tutorials for students and researchers interested in applying AI techniques to scientific and engineering challenges. It provides foundational knowledge and hands-on experience with cutting-edge AI methods relevant to fields like physics-informed neural networks, neural operators, and graph neural networks.

How It Works

The course covers a range of AI methodologies tailored for scientific applications, including physics-informed neural networks (PINNs) for solving partial differential equations (PDEs), neural operators (like Fourier Neural Operators and DeepONets) for efficient surrogate modeling, and graph neural networks for molecular design and computational biology. These techniques are presented as alternatives or complements to traditional scientific workflows, aiming to improve efficiency and performance across diverse domains such as fluid dynamics and medical physics.

Quick Start & Requirements

The repository contains links to lecture recordings and detailed tutorials covering topics from basic PyTorch function approximation to advanced concepts like PINNs, operator learning, graph neural networks, and diffusion models. Specific datasets, such as the Allen-Cahn dataset, are linked for practical exercises.

Highlighted Details

• Covers a broad spectrum of AI techniques relevant to scientific discovery and engineering.
• Includes practical tutorials on PyTorch, PINNs, Neural Operators, GNNs, and diffusion models.
• Demonstrates applications across fluid dynamics, wave physics, medical physics, molecular design, and computational biology.
• Features new tutorials for Fall 2024, including Time Dependent CNO and Transfer Learning for PDEs.

Maintenance & Community

The course is led by Prof. Dr. Siddhartha Mishra and Dr. Ben Moseley, with assistance from Bogdan Raonic and Victor Armegioiu. Links to lecture recordings and the official course page are provided.

Licensing & Compatibility

The repository itself does not specify a license. The content is presented for educational purposes related to the ETH Zurich course.

Limitations & Caveats

This repository is primarily an educational resource and project page, not a deployable software library. The practical application of the techniques requires significant background knowledge in AI and scientific computing, along with the necessary computational resources for training models.