「Comprehensive analysis of hyper-ordered structures based on mathematics and informatics」
（Tohoku University, Professor, Statistical machine learning）
(Nara Institute of Science and Technology, Professor, Data analysis on atomic resolution holography)
(Okayama University, Professor, Analytic geometry, Persistent homology)
(Kyoto University, Professor, Applied mathematics, Persistent homology)
Development of informatics methods for identifying structures (atomic configurations) and for predicting material properties of hyper-ordered materials. These methods can contribute for rational material designs of hyper-ordered materials.
For material measurement data such as atomic resolution holography and angstrom beam electron diffraction, we will develop structure identification methods based on machine learning and reverse Monte-Carlo (RMC) method. Our group will also develop novel descriptors for structure orders in hyper-ordered materials based on geometry and topology. Developed descriptors will be used to develop machine learning models to predict materials properties. Using our informatics methods, we will further develop machine learning methods for rational material and process design.
On structure data analysis, we will collaborate with group A02-1 and A02-2. For developing structure descriptors, simulation results provided from group A03-1 will be used. Content 3 is collaborative works with group A01-1 and A01-2 for accelerating new material development.