"Neutron and X-ray Diffraction Reveal the Limits of Long-Range Machine Learning Potentials for Medium-Range Order in Silica Glass"

This dataset contains selected liquid and amorphous silica trajectories, trained MACE-SR and MACE-LR model files, simulation scripts, and processed structural-analysis data associated with the manuscript “Neutron and X-ray Diffraction Reveal the Limits of Long-Range Machine Learning Potentials for Medium-Range Order in Silica Glass.”

The dataset is organized into separate archives for ease of use: trajectories.zip, data.zip, scripts.zip, and MACE_Models.zip. The trajectory archive contains liquid and amorphous silica trajectories generated using both short-range and long-range MACE models, including amorphous quench trajectories at multiple cooling rates. The processed data archive contains RDFs, bond-angle distributions, ring statistics, and topological data analysis outputs, including persistent-homology and accumulated-persistence-function data. The scripts archive contains the melt-quench simulation workflow, and the model archive contains the trained MACE model files and associated reference datasets.

These files are provided to support reproduction of the structural analyses reported in the manuscript and Supporting Information, including radial distribution functions, bond-angle distributions, ring-size distributions, and persistent-homology-based analysis of medium-range order.