Berlin 2018 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

MA: Fachverband Magnetismus

MA 13: Focus Session: Magnetism in Materials Science: Thermodynamics, Kinetics and Defects II (joint session MM/MA)

MA 13.7: Talk

Monday, March 12, 2018, 18:15–18:30, TC 010

Machine Learning assisted Heisenberg model for systems with ill-defined pairwise magnetic interactions — •Osamu Waseda, Omkar Hegde, and Tilmann Hickel — MPIE

Magnetic interactions are important for the stability of structural phases as well as for various thermophysical effects such as magnetocalorics. In order to determine their free energy contribution in Fe-based materials, the Heisenberg model has been used as a handy method for decades. Despite its simplicity, there is little experience with the application of this model to systems containing various types of atoms and/or structural defects, as their interaction parameters cannot be defined straightforwardly. In this study, data sets for Fe-Mn systems containing structural defects are created from spin-polarized DFT calculations. They are then translated into the Heisenberg parameters via Ridge regression. Finally, the contribution of the magnetic interactions to the specific heat is determined through Monte Carlo simulations.