Data for paper "Automated Structure Discovery for Scanning Tunneling Microscopy"

Contents of the dataset:

band.h5 -- keys are molecule indices, each molecule has the following keys:

eigs: KS eigenvalues for each state

coefs: KS eigenvectors for each basis set

xyz: atomic positions

Z: atomic species

qs: mulliken point charges

rotations_210611.pickle -- keys train/val/test

Each set is a dict containing id -- rotation pairs

rotations are 3x3 numpy arrays

disks.pt -- a pretrained model for Atomic Disks predictions