Abstract
Material Informatics is one of the central subject in materials science today. Coercivity of magnetic material is an important property determining power generation efficiency of motors; however, the analytical methodology has not been perfectly developed.
Here, we applied “Persistent Homology” to the magnetic domain structure to describe macroscopic magnetic properties. Persistent Homology is a powerful methodology for quantitative evaluation of topological feature in structural data. Moreover, it provides us the inverse analysis from macroscopic property to structural data in the combination with machine learning. We demonstrate Persistent Homology analysis on magnetic domain structure, and examine the validity as a descriptor.
