Dataset: CSP-generated crystal structures of 1,000+ rigid organic molecules

This dataset supports the publication:
AUTHORS: Christopher R. Taylor, Patrick W. V. Butler, Graeme M. Day
TITLE: Predictive crystallography at scale: mapping, validating, and learning from 1,000 crystal energy landscapes
JOURNAL: Faraday Discussions

A consolidated dataset of crystal structure predictions (CSPs) for 1007 unique rigid, organic molecules with observed crystal structures in the Cambridge Structural Database (CSD). Each CSP is described by a "landscape" of hypothetical crystal structures, ranked in terms of their lattice energy; this dataset includes both the crystal structures themselves and their energy rankings.

This dataset also includes two machine-learning-derived models to improve the energy ranking of crystal structures on their respective landscapes; one a committee neural-network potential (NNP) to correct energies of fixed structures, the other a message-passing neural-network (MACE) model used to re-optimise particularly difficult crystal structures.