Computational data for "Machine-Learnt Fragment-Based Energies for Crystal Structure Prediction"

Crystal structure prediction datasets and calculated energies, supporting the publication "Machine-Learnt Fragment-Based Energies for Crystal Structure Prediction" The file CSP_cif_files.zip contains all crystal structures generated for the molecules in the publication (3,4-cyclobutylfuran, adamantane, adenine, formamide, maleic hydrazide, naphthalene, oxalic acid, tetrolic acid, triazine, urazole), within a 20 kJ/mol lattice energy window from the global minimum, separately for each molecule. The spreadsheet energy_data.xlsx contains the calculated lattice energies for all predicted crystal structures using the force field (FIT+DMA) and three fragment-corrected energy models.