Computational data for: Multi-fidelity Statistical Machine Learning for Molecular Crystal Structure Prediction

Abstract

Crystal structure prediction dataset for molecules included in the publication Multi-fidelity Statistical Machine Learning for Molecular Crystal Structure Prediction (2020). Crystal structure files (in CIF format) for the three molecules: oxalic acid, urazole and maleic hydrazide. All predicted crystal structures for each molecule within 25 kJ/mol of the lowest energy structure (as calculated with the FIT+DMA force field) are included. The file name for each structure represents its structure key and is generated during the crystal structure prediction procedure. A file per molecule containing calculated energies for each crystal structure (identified by its structure key) at the force field level (FIT+DMA), PBE with D3 dispersion correction and PBE0 with D3 dispersion correction and geometric counterpoise correction.