Caltech's O-kRystal cuts PBE band energy MAE 68.7%, learning beyond-DFT corrections
AnimaAnandkumar · x · 2026-10-18
Anima Anandkumar's Caltech team introduces O-kRystal, a periodic E(3)-equivariant graph neural network that learns corrections to individual electronic states beyond standard DFT.
- Method: it uses orbital information from cheap PBE calculations in both real and reciprocal space, combining local electronic interactions with crystal momentum dependence in one model.
- Three firsts: HSE06 band predictions across nearly 6,000 chemically diverse 3D materials; coverage of metals/insulators, magnetic/nonmagnetic systems with spin resolution; and, trained separately, G0W0 quasiparticle prediction in 2D materials.
- Trained only at high-symmetry points, it keeps comparable accuracy along unseen band paths and reconstructs densities of states.
- On a strongly correlated materials benchmark, it reduces PBE band energy MAE by 68.7%; for AFM-II MnO (a canonical Mott insulator) it captures distinct Mn 3d and O 2p corrections.
The framework points toward experimental-accuracy electronic structure and large-scale quantum material discovery.
Related event: Caltech's O-kRystal Learns Band Structure Corrections Beyond DFT Accuracy(2 posts)→
More from Research
- Claude Opus simulated Hox gene networks to generate lizard and snake skeletons — repligate · 2026-10-18
- OpenEnv Arena lets you build RL environments without GPUs, one training run per 24h — ben_burtenshaw · 2026-10-18
- French math PhD students despair: a €20 ChatGPT could solve their thesis problems — FlorianGallwitz · 2026-10-18
- 'Idea Basins': independent runs of the same task plateau at different scores compute can't close — menhguin · 2026-10-18
- NVIDIA researcher counters linear attention debate: 8B hybrid beat Transformer 7 months before MiniMax-01 — ctnzr · 2026-10-18
- ERRORPROBE: automatically pinpointing which agent caused a multi-agent failure — burkov · 2026-10-18