Research data supporting "Predicting novel superconducting hydrides using machine learning approaches"

Crystal structures of the materials for which critical temperatures were calculated in the paper "Predicting novel superconducting hydrides using machine learning approaches" (https://arxiv.org/abs/2001.09852). These crystal structures were generated by selecting low-enthalpy candidates from a random structure search, and performing a geometry optimization at the pressure(s) of interest (the parameters for which are included in each file). The data consists of a set of crystal structure files are named with the following format: a_b_c_d_e_kpts_scf.in where a = the stoichiometry of the material b = the space group of the crystal c = the number of formula units per primitive cell d = pressure at which relaxed e = "primary", or "aux" corresponding to the two different k-point grids used These files are human-readable and contain the crystal lattice specification under the section CELL_PARAMETERS and the atomic positions within the lattice under the ATOMIC_POSITIONS, as well as the various named parameters used in the density functional theory calculations. They may also be read by the quantum-espresso software (https://www.quantum-espresso.org/) or converted to many common crystal-structure formats using the c2x software (https://www.c2x.org.uk/).