This repository serves as a curated, ranked list of "awesome" open-source projects in atomistic machine learning (AML). It aims to provide researchers and practitioners with a comprehensive overview and quality assessment of AML tools, datasets, and methodologies, focusing primarily on simulation data and materials science.

How It Works

The list is automatically generated and updated, ranking projects based on a composite "project-quality score" derived from GitHub metrics (stars, forks, contributors, issues) and package manager data (downloads). Projects are categorized, and indicators are provided for activity status (active, inactive, dead), trending, and recency.

Quick Start & Requirements

• Installation: Projects are typically installed via pip or conda, or cloned using git clone. Specific commands are provided for many listed projects.
• Dependencies: Varies by project; common dependencies include Python, PyTorch, TensorFlow, JAX, and specific scientific libraries. Some projects may require GPU acceleration or specific CUDA versions.
• Resources: Setup time and resource requirements vary significantly per project. Links to official documentation, demos, and GitHub repositories are provided for each entry.

Highlighted Details

• Comprehensive Categorization: Over 23 categories cover diverse AML subfields, from Interatomic Potentials and Representation Learning to Generative Models and Explainable AI.
• Project Quality Scoring: A unique scoring system helps users quickly identify high-impact and well-maintained projects.
• Regular Updates: The list is regularly updated, ensuring relevance and inclusion of new developments in the rapidly evolving AML field.
• Community Driven: Contributions and updates are encouraged via GitHub issues and pull requests, fostering a collaborative environment.

Maintenance & Community

The project is maintained by the JuDFTteam. Community interaction and contributions are facilitated through GitHub issues and pull requests.

Licensing & Compatibility

Licenses vary by project, including MIT, BSD, GPL, LGPL, Apache-2.0, and custom licenses. Users must check individual project licenses for compatibility with commercial or closed-source applications.

Limitations & Caveats

The list's primary focus is on simulation data and materials science, potentially underrepresenting experimental data or drug design applications. Some listed projects may be inactive or deprecated, as indicated by specific markers.