graph-pes (University of Oxford)
github.com/vldgroup/graph-pesUnified framework for building, training, and deploying graph-based machine-learned interatomic potentials: a single graph-pes-train pipeline trains many architectures (PaiNN, NequIP, MACE, and more) from scratch or fine-tunes existing foundation models, with drivers for structure optimisation and MD via ASE, LAMMPS, and torch-sim; pip/conda-installable with a clean, documented API for prototyping new MLIP methods (Deringer group, 131+ stars, MIT, v1.0.2 with Zenodo DOI, 2026)
Sourced from
- Awesome AI for Science — github.com/vldgroup/graph-pes
Related resources
Meta's comprehensive ML ecosystem for materials/chemistry with 118M+ DFT calculations, EquiformerV2 models achieving top Matbench Discovery performance
Deep learning package for many-body potential energy representation and molecular dynamics, achieving quantum-mechanical accuracy with classical MD efficiency (DeepModeling, Gordon Bell Prize 2020, 1.9k+ stars)
Python Materials Genomics: robust materials analysis library defining classes for structures and molecules with support for many electronic structure codes; foundational toolkit powering the Materials Project (Berkeley Lab, 1.8K+ stars)
Diffusion-based generative model for inorganic materials design, steering generation by chemistry, symmetry, bulk modulus, band gap, or magnetic properties, 2× more likely to produce stable novel structures than prior methods, experimentally validated with synthesized TaCr₂O₆ (Microsoft, Nature 2025)
Machine learning interatomic potentials
DeepMind's graph neural network for materials exploration, discovering 2.2M new crystal structures (380K most stable) equivalent to 800 years of traditional research, with 520K+ materials dataset open-sourced (Nature 2023)