HES-ML: Training Datasets and Deep Potentials for Silica Gas-Surface Interactions

This repository contains the Deep Potential (DP) models and HDF5 training datasets generated via the HES-ML (hybrid enhanced sampling-machine learning) framework. The framework integrates Deep Potential Molecular Dynamics (DeePMD) with dual-mode On-the-fly Probability Enhanced Sampling (OPES) to model extreme gas-surface interactions for Silica systems. The data covers bulk melting, surface reconstruction, and complex oxygen adsorption/recombination reaction pathways.