Dr. Runhai Ouyang

Materials Genome Institute, Shanghai University

The materials-genome initiative has fostered high-throughput calculations and experiments, leading to large amount of materials data available in literature and databases. Identifying the trends and correlations hidden in the materials data and thereby accelerating materials discovery is the core of the emerging fourth paradigm of science for materials science: data-driven material discovery. In this regard, efficient data-driven approaches for descriptor identification are crucial, and many methods falling under the umbrella name of (big-) data analytics (e.g. data mining, machine learning, compressed sensing, etc.) have being developed and applied to the wealth of materials-science data. In this talk, Ouyang will introduce their recently developed novel data-driven method SISSO based on compressed sensing theory for identifying low-dimensional descriptors for materials’ properties and functions. Then he will show several successful applications of SISSO in materials science to demonstrate the efficiency, e.g. new tolerance factor for predicting the geometry stability of perovskite, physical descriptor for predicting the Gibbs energy of crystalline solids, materials map for predicting topological insulators. In addition, SISSO is expected to be an efficient tool for big-data catalysis.