Predicting 3D Magnetic Topological Insulators and Semimetals with Machine Learning

The search for magnetic topological insulators and semimetals is important for the application of spintronic devices. In the past, identifying topological insulators and semimetals has required time-intensive Density Functional Theory calculations followed by a Hamiltonian model using Maximally Localized Wannier Functions. We use a machine learning (ML) model to identify topological insulators and semimetals, thus allowing for a quicker identification process. The materials of interest in this project are A_xB_yC_z compounds where A = Cobalt, Chromium, Europium, Nickel, Vanadium, B = Bismuth, Antimony, and C = Tellurium or Selenium. The ML model for topological insulators is based on atomic properties while the model for semimetals also uses information on the crystal structure. Several different neural net models were evaluated for the semimetal model. The accuracy of these models was checked with ab initio calculations for materials in the A_xB_yC_z family leading to a number of previously unidentified materials being confirmed as magnetic topological insulators and semimetals.