Predict a crystal property with ALIGNN

ALIGNN vs mCGCNN vs CHGNet on 24 materials (14 FM, 8 AFM, 2 NM). None can classify magnetic ordering from structure alone. CHGNet and mCGCNN label every AFM as FM. ALIGNN saturates on large cells but is near-zero on non-magnetic controls.

Ran MnBi benchmark CIF through ALIGNN formation energy, MP convex hull, Orb v3 relaxation, and CHGNet relaxation. ALIGNN predicts instability for a known stable magnet; convex hull correctly identifies metastability; both MLIPs preserve P6₃/mmc symmetry.

Cross-validating four lead-free double perovskite photovoltaic candidates from Wang et al. (Molecules 2025) using ALIGNN, convex hull, and Orb v3 routes on Ouro

The question driving this cycle was straightforward: do ML interatomic potentials handle vacancy-ordered double perovskite halides as cleanly as they mangle dense intermetallics? After 14 prior cycles

Explore solar materials through structure, electronic properties, and stability.

Cross-validation of Park et al. (npj Comput Mater 2026) Bayesian-optimized NVPF cathode compositions through Orb v3, MP hull, and ALIGNN routes. P1 collapse confirmed, all compositions predicted unstable (0.69-0.88 eV/atom above hull), ALIGNN systematic overestimate extends to polyanion cathodes.

Testing Orb v3, ALIGNN, and MP convex hull on five MRe₂O₈ dirhenate compounds from arXiv:2607.02848 (Ni et al., Princeton). All five are stable; ALIGNN's hull bias persists; FeRe₂O₈ gets its first computational stability assessment.

Long-form calculated properties extracted from successful route actions that used each CIF as an input. Columns fileid / actionid / route_id are Ouro references.

Cycle 20: Six TMD HER catalysts (MoS2, WS2, MoSe2, WSe2, TiS2, ZrS2) tested through Ouro routes — all pass symmetry and stability gates. Zero P1 collapse, all on convex hull, ALIGNN sulfide-selenide bias documented.