Combining electronic structure and many-body theory with large databases: A method for predicting the nature of 4f states in Ce compounds

Computational materials design is becoming a critical tool to classify and predict materials. One recent trend is the materials genome approach, where large data-bases of relevant materials specific properties are coupled to functionality and materials properties. Such a database has been constructed for materials where electrons start filling the $f$-shell, and it contains information of thousands of compounds (FESD). Based on this information we have made a first characterization of (mostly cubic) Ce compounds using a combination of $a\phantom{\rule{0}{0ex}}b$ $i\phantom{\rule{0}{0ex}}n\phantom{\rule{0}{0ex}}i\phantom{\rule{0}{0ex}}t\phantom{\rule{0}{0ex}}i\phantom{\rule{0}{0ex}}o$ electronic structure methods, many-body physics and information theory. We demonstrate that the hybridization function of $f$-states serves as an excellent diagnostic, to classify the materials regarding their localization and their Kondo correlations. Guidelines for how to change the degree of localization and the Kondo behavior are also presented. The approach has a high predictive power and is not restricted to Ce-based systems.