All presented data were generated based on trajectories of ab initio molecular dynamic (MD) simulations using VASP code. Twelve molecules were simulated within NPT ensemble.

sorbitol.png contains schematic representation of sorbitol molecule with atom numbers used in data arrays indexing schemes.

For each temperature (80, 40, 200, 260, 320 and 380 K) there are four files:

T_POSCAR is VASP style POSCAR file, containing lattice vectors, atoms positions and velocities at the end of MD run.

T_desc.npz is a python dictionary containing arrays of structural MD data that can be accessed wit appropriate keys: ‘doc’: O-C pair distance, ‘doh’: O-H pair distance, ‘vcoh’: C-O-H triplet bend, ‘da’: H-O-C-C quartet dihedral stretch, ‘dh_o’: H--O distance for hydrogen bonds, ‘do_o’: O—O distance for hydrogen bonds, ‘voh_o’: O-H—O triplet bend, ‘sd’: binary data for each hydrogen bond with True value for intra-molecular bonds and False otherwise, ‘hmap’: matrix containing molecule and atom number for each hydrogen bond at each time_step. Each matrix contains data indexed according to the [time_step, molecule, atom] scheme.

T_rear_ccoh is a python array with dihedral rearrangement data for each hydroxyl group hydrogen atom, indexed according to the [time_step, molecule*6 + atom] scheme.

T_rear_h is a python array with spatial rearrangement data for each hydrogen atom, indexed according to the [time_step, molecule*14 + atom] scheme.