Research data for "Crystal structure identification with 3D convolutional neural networks with application to high-pressure phase transitions in SiO2"

This dataset supports the paper "Crystal structure identification with 3D convolutional neural networks with application to high-pressure phase transitions in SiO2". The following files are provided: -The training database for the simple (artificial and MD) and the SiO2 structures --> The training data is provided in two different formats. In the "simple_training_dump" and "SiO2_training_dump" files, the dump files from the MD trajectories are provided. In the "simple_training_extracted" and "SiO2_training_extracted" files 1,000,000 extracted atomic environments in a numpy format are stored. -The holdout dataset for the simple structures -The snapshots of the SiO2 shock simulation