Multiscale models in condensed matter

Abstract

My research activities are focused on multiscale phenomena in condensed matter. Firstly, I have studied in collaboration with A. Crisanti and L. Leuzzi, the phenomenon of reversible inverse transition (IT), occurring between a solid and a liquid in the inverse order relation relatively to standard transitions. In particular we have studied, through Exchange Monte Carlo Simulaiton, the static of the so-called Blume-Capel (BC) model with quenched disorder (BCrandom) in three dimensions: an Ising-like spin glass model with spin-1 variables. From the dynamical poitn of view, I am investigating with A. Crisanti and L. Leuzzi the behaviour of a class of p-spin model (i. e., the so-called (p+s)-spin model). The (p+s)-model displays the feature of having both usual “one-step” RSB solution, known to reproduce all basic properties of structural glasses, and a physically consistent “two-step” solution. Along certain cooling paths in the phase diagram, determined by analysis of the static themodynamic properties, approaching the tricritical point MCT applied to the s+p model yields a time-correlation function with two plateaus at different correlation values. Under the supervision of A. Di Carlo we are investigating some topics about the coupling of continuum and atomistic mechanics. The Andersen-Parrinello-Rahman method is based on an extended Lagrangian allowing the MD cell to change both volume and shape along the simulation, its dynamics being governed by an externally applied stress, as well as by the internal particle dynamics. The APR method is well established and widely used. However, some foundational issues remained unexplored until recently, since the Lagrangian introduced by Andersen and generalized by Parrinello and Rahman has always been considered just as an expedient trick for generating the desired particle statistics. We assume a different standpoint: to us, APR-like Lagrangians embody the coupling between atomistic and continuum degrees of freedom. So, we are interested in the dynamics of the deforming computational cell per se, wishing to identify it with the body element of a Cauchy continuum. On this basis, we plan to construct atomistically informed approximations to a continuum by means of an array of interacting APR-like cells.