Hamiltonian Learning - Small dataset of Ethane and QM7

A small dataset of rattled ethane and selected structures of the QM7 dataset (New J. Phys. 15, 095003 (2013)).The dataset contains the structures as xyz files, along with the Hamiltonian (Fock) matrices, the overlap matrices, the eigenvalues, the dipole moments, the polarization, and the orbitals present for the relevant basis set (STO-3G and dev-TZVP) computed with PySCF. All these files are saved in hickle format. This dataset is used for an atomistic cookbook example called Hamiltonian Learning for Molecules with Indirect Targets to learn indirectly the Hamiltonian by targetting properties derived from the Hamiltonian (eigenvalues, dipole moments, and polarization), analogous to the results in this preprint: https://arxiv.org/abs/2504.01187.