About the Platform
This project explores cutting-edge applications of advanced AI algorithms in the research and development of multi-principal-element high-entropy alloys. By deeply modeling the complex mapping between compositional ratios, microstructural features, and macroscopic physical and thermodynamic properties, it builds an intelligent auxiliary computing tool with strong generalization capability — significantly reducing trial-and-error cost in new-materials development and providing a data-driven scientific foundation and technical support for the design of future ultra-high-performance alloys.
Special Acknowledgements
Throughout the construction of high-dimensional feature engineering and the training of algorithmic models, this project has deeply upheld the collaborative spirit of the global materials-science community. We hereby express our sincere gratitude to NIST (National Institute of Standards and Technology), LBNL (Lawrence Berkeley National Laboratory), and NIMS (National Institute for Materials Science) for the invaluable support provided by their internationally authoritative materials-science databases. The current website is built on the Materials Project database, and training will draw on all available partner materials databases.
Materials Project
Established by Lawrence Berkeley National Laboratory (LBNL), the Materials Project is one of the most renowned databases in materials science. It has provided us with extensive crystallographic, thermodynamic, electromagnetic, and mechanical data on multi-component and high-entropy alloys.
Visit the MP Database
NIST JARVIS
Founded by the National Institute of Standards and Technology (NIST). Its open, large-scale, high-precision mechanical and elastic-tensor DFT data has built an essential high-dimensional feature space and benchmark reference for exploring the intrinsic mechanical properties of high-entropy alloys.
Visit the NIST JARVIS Database
NIMS DICE
Coordinated by the National Institute for Materials Science (NIMS), Japan. Its collection of real experimental property data for complex multi-component systems bridges the gap between theoretical computation and engineering practice, providing irreplaceable measurement-grade support for the generalization of machine learning under realistic conditions.
Visit the NIMS DICE DatabaseVersion Information
The database version used by this website may lag behind the latest release of the Materials Project database; refer to the Materials Project version notes for specifics. When obtaining or using data, we recommend paying attention to the last-updated timestamps. Because incremental synchronization is used, some historical data may remain in the system, and we may also perform occasional full refreshes of already-synchronized data as circumstances require. Data provided by NIST and NIMS requires authentication before it may be accessed in controlled systems, and such data is governed by its own separate terms and versions.
