| Abstract Scope |
Through the ARPA-E ULTIMATE program, the team at Penn State developed a data ecosystem ULTERA designed by Krajewski(1) and used it for cGAN inverse design of refractory HEMs(2). ULTERA consists of four automated database-driven flow loops, i.e., literature, design, validation, and prediction. With 550+ unique DOIs and 6800+ unique experimental data points, ULTERA features a robust data curation infrastructure with a set of data validation, processing, and aggregation tools. Unique tools include PyQAlloy(3) for detecting data abnormalities and nimCSO(4) for optimizing design space. In this presentation, features of ULTERA will be discussed in the framework of our AI-driven high throughput prediction and modeling of materials properties(5), including materials design(6) and efficient generation of grids and traversal graphs for functionally graded materials(7).
(1)https://ultera.org;
(2)J. Mater. Informatics 1, 3 (2021);
(3)https://pyqalloy.readthedocs.io/;
(4)ArXiv 2403.02340 (2024);
(5)https://github.com/PhasesResearchLab/SoftwareProjects;
(6)J. Mater. Res. 38, 4107–4117 (2023);
(7)ArXiv 2402.03528 (2024). |