Software & data.
We develop open materials databases, machine-learning models, and computational workflows that transform quantum and atomistic information into reproducible material, process, and system-level predictions.
{{ sec.title }}
{{ t.name }}
{{ t.kind }}{{ t.desc }}
A fast machine learning model that predicts DFT-quality partial atomic charges (DDEC06, CM5, Bader, and REPEAT) for MOFs and COFs in seconds per structure (vs. hours to days for DFT), with mean absolute errors of ~0.01 e against DFT-derived charges.
{{ pp.title }} · {{ pp.authors }}, {{ pp.journal }}, {{ pp.biblio }}