Finding two-level systems in glasses through machine learning

Authors:
  • SCSimone Ciarella
    Simone Ciarella
  • DKDmytro Khomenko
    Dmytro Khomenko
  • LBLudovic Berthier
    Ludovic Berthier
  • FMFelix C. Mocanu
    Felix C. Mocanu
  • DRDavid R. Reichman
    David R. Reichman
  • CSCamille Scalliet
    Camille Scalliet
  • FZFrancesco Zamponi
    Francesco Zamponi

Abstract
Two-level systems (TLS) are rare quantum tunneling defects which govern the physics of glasses at very low temperature. Because of their extremely low density, it is very hard to directly identify them in computer simulations of model glasses. We introduce a machine learning approach to efficiently explore the potential energy landscape of glass models and identify two-level tunneling defects. We design an algorithm that is able to rapidly predict the quantum splitting between any two amorphous configurations produced by classical simulations. This in turn allows us to shift the computational effort towards the collection and identification of a larger number of TLS, rather than the useless characterization of non-tunneling defects which are much more abundant. Finally, we interpret our machine learning model to understand how TLS are identified and characterized, thus giving physical insight into the features responsible for their presence.