Presentation Information
[9p-PA2-20]PHYSBO-Based Search System and Theoretical Evaluation of Novel Superconducting Materials
〇(M2)Yuma Shoji1, Hirofumi Sakakibara1,2 (1.Fac. of Eng., Tottori Univ., 2.AMES, Tottori Univ.)
Keywords:
superconductivity,Machine Learning,PHYSBO
The practical application of high-temperature superconductors is hindered by their poor mechanical properties and the reliance on empirical trial-and-error for discovering new materials. To address these issues, this study aims to build an automated exploration system using machine learning to efficiently discover novel superconductors with both high transition temperatures and excellent mechanical properties.
In this poster presentation, we report on the initial phase of this project: the development of a system that uses Bayesian optimization to learn structural parameters of known cuprate superconductors and efficiently narrow down candidates from a vast material space. Furthermore, we evaluate the superconducting properties of the extracted candidates using first-principles calculations and the FLEX approximation. Detailed benchmark results and the correlation between crystal structure and superconductivity will be presented.
In this poster presentation, we report on the initial phase of this project: the development of a system that uses Bayesian optimization to learn structural parameters of known cuprate superconductors and efficiently narrow down candidates from a vast material space. Furthermore, we evaluate the superconducting properties of the extracted candidates using first-principles calculations and the FLEX approximation. Detailed benchmark results and the correlation between crystal structure and superconductivity will be presented.