Discovery pipeline for magnetic materials: unified dataset, multitask crystal GNN screening, GNoME candidate ranking, and DFT validation of rare-earth-free permanent-magnet candidates.
-
Updated
Apr 24, 2026 - Python
Discovery pipeline for magnetic materials: unified dataset, multitask crystal GNN screening, GNoME candidate ranking, and DFT validation of rare-earth-free permanent-magnet candidates.
Physics-grounded screen of DeepMind GNoME (554,054 structures) for permanent- and soft-magnet candidates, with noncollinear spin-orbit DFT anisotropy validation
Computational screen of 139,894 ML-discovered DFT-relaxed crystals for rare-earth-free permanent magnets. Independently rediscovers a-Fe16N2 as the top-ranked candidate.
Reproduces every figure of the paper 'Beyond Stoner-Wohlfarth: machine-learning models and symbolic regression of hard-magnet properties' from committed data — one script per figure
Build it. Break it. Beat it. Public evidence for the ARLO-S1 AI-assisted engineering challenge.
To associate your repository with the permanent-magnets topic, visit your repo's landing page and select "manage topics."