Artificial Intelligence For Use In Atomistic Kinetic Monte Carlo Simulations

Flyura Djurabekova, Roberto Domingos, Gennaro Cerchiara, Nicolas Castin, Lorenzo Malerba, Abderrahim Al Mazouzi

    Research outputpeer-review


    A new approach to atomistic kinetic Monte Carlo (AKMC) simulations based on the determination of vacancy migration barriers as functions of the local atomic environment has been developed, with a view to provide a better description of the kinetic path followed by the system through diffusion of solute atoms in the alloy via vacancy mechanism. Tabulated values of barriers versus local atomic configurations (LAC) including atoms up to 2nd nearest neighbour shell, obtained by molecular dynamics (MD) techniques, have been used to train an artificial intelligence (AI) system to recognize the LACs and predict the barriers accordingly. Here some details on the method and preliminary results are presented and briefly discussed.
    Original languageEnglish
    Title of host publicationMMM*** Third International Conference Multiscale Materials Modeling
    Place of PublicationFreiburg, Germany
    StatePublished - Sep 2006
    Event3rd Intl Conf on Multiscale Materials Modeling - Fraunhoffer Institute for Mechanics of Materials - University of Freiburg, Freiburg
    Duration: 18 Sep 200622 Sep 2006


    Conference3rd Intl Conf on Multiscale Materials Modeling

    Cite this