Research Director

Hideyuki Ishi

Graduate School of Science,Osaka City University
Professor

Outline

Symmetric matrices of huge size with many zero entries, called sparse symmetric matrices, are nowadays studied actively in the context of artificial intelligence and data science. One of the efficient approach to such matrices is the graphical model method. We succeed to introduce a general structure of symmetric matrices which unifies the theroy of graphical model and the theory of homogeneous cones. Based on this discovery, we will evaluate, improve and innovate various algorithm in matrix calculation via the idea of homogeneous cones. In particular, we expect essential contribution in optimization problems such as the convex cone programming.