Data associated with the paper "Atomistic Mechanisms of Pressure-Induced Phase Transitions in Heat-Storage Material Ti3O5"

The data includes the crystal structures of all predicted stable and metastable phases, training and validation datasets, and the developed machine learning potential. The data includes all structures visualized in the manuscript and used for the data in the manuscript plots and supporting information, the machine learned potential, the training and validation datasets, the modified Lammps code snippet to print the force components resulting from applied repulsive potentials and the script used to analyze a trajectory produced by the patched Lammps code to calculate the pressure.