The structures and properties of the newly predicted compounds, and the artificial neural network machine learning interatomic potentials developed in this project, are documented in https://ml-material.physics.iastate.edu/database. The data will be made available for public to download and use when the papers reporting these results are published.
The Database provides a large array of datasets for novel compounds, with focus on functional materials containing compounds with three or more elements. Our database is specifically designed for data sciences and machine learning modeling. Available datasets include (i) crystallographic data, (ii) ML interatomic potentials trained with DeepMD package. Users can download these data in various file formats. Our data can be visualized by using our web applications.
The open-source software developed in this project is available on GitHub:
We also provide relevant software used in our ML-guided Materials Discovery Framework.