Researcher, ITMO University (Saint-Petersburg, Russia)
04 July 2018, 15:00-16:30
SkolTech, TPOC-4, Nobel str. 1, Red Building, 3rd floor, Room 351
ABSTRACT. Recently, a deep connection between physics of interacting quantum systems and machine learning technique has been pointed out. In fact, both of the fileds deal with systems with an extremely large number of degrees of freedom. In physics this problem is solved within coarse-grained modeling which reduces a complex many-body system to a more simplified one. Whereas in machine learning routine one typically employs dimensional reduction in the data space. In this talk we discuss how to use both supervised and unsupervised machine learning to classify phases of two-dimensional Heisenberg magnet and bcc-Fe.
Short BIO. Dmitry Yudin studied physics at Moscow Institute of Physics and Technology (Moscow, Russia) and graduated with a PhD examination at Uppsala University (Uppsala, Sweden). During his PhD he was jointly supervised by Prof. Olle Eriksson (Uppsala University, Sweden) and Prof. Mikhail Katsnelson (Radboud University Nijmegen, the Netherlands) with the strong emphasis on studying strongly correlated electronic systems, magnetism and spin-dependent phenomena. He contributed to the development of advanced numerical technique, namely dual fermion approach, which allows to generalize a very successful dynamical mean-field theory to include spacial correlations. Another line of research activity he was pursuing at the time is studying of topologically nontrivial spin textures on exotic lattices, e.g. a skyrmion propagating along the edge of kagome lattice. He then took a position as a Research Fellow at Nanyang Technological University (Singapore) and subsequently at the ITMO University (Saint Petersburg, Russia) where he is still employed as a Researcher. He is currently active in developing a unified framework for studying spin-transfer and spin-orbit torques for antiferromagnetic spintronics.