QM7 Decomposed Data

This data is in support of our article 'Decomposing Chemical Space: Applications to the Machine Learning of Atomic Energies' (arXiv:2212.09489). An example python script is provided with more details about the PySCF and DECODENSE settings. DATA The data is saved in Numpy .npz format. The files are named as b3lyp_{basis_set}_qm7_{decomposition}_atomization.npz The keys in each file correspond to: 'mol_idx': Molecule index corresponding to the molecule index in original QM7 dataset. 'N' : Number of atoms in the molecule 'Z' : Nuclear charges 'R' : Coordinates (ångstrom) 'E' : Total atomization energy (kcal/mol) 'Ea' : Atomization energy per atom (kcal/mol)