Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 9: Focus Session: Frontiers of Electronic-Structure Theory I (joint session O/HL)
O 9.3: Topical Talk
Monday, March 27, 2023, 11:00–11:30, TRE Ma
Large-scale machine-learning assisted discovery and characterization of materials — •Miguel Alexandre Lopes Marques — Institut für Physik Martin-Luther-Universität Halle-Wittenberg, Halle (Saale), Germany
In this talk we discuss our recent attempts to discover, characterize, and understand inorganic compounds using ab initio approaches accelerated by machine learning. We start by motivating why the search for new materials is nowadays one of the most pressing technological problems. Then we summarize our recent work in using crystal-graph attention neural networks for the prediction of materials properties. To train these networks, we curated a dataset of over 2 million density-functional calculations with consistent calculation parameters. Combining the data and the newly developed networks we have already scanned more than two thousand prototypes spanning a space of more than one billion materials and identified tens of thousands of theoretically stable compounds. We then discuss how simple, interpretable machine learning approaches can be used to understand complex material properties, such as the transition temperature of superconductors. Finally, we speculate which role machine learning will have in the future of materials science.