Metastable states, quasi-stationary distributions and soft measures (Q271842)

In some systems, for example freezing fog made of small droplets, one can observe the following: {\parindent=0.8cm \begin{itemize}\item[(i)] the presence of only one thermodynamic phase -- liquid phase, \item[(ii)] the initial state of a system lasts for a long time (a long time is needed to get out -- fog is visible for a long time until collision with ground or trees), \item[(iii)] if the system has gotten out of this state, it is unlikely to return -- frozen fog will typically not return to liquid state before pressure or temperature has changed. \end{itemize}} Such systems are characterized as metastable thermodynamic states with the above-mentioned features. They can be modelled by the Lebowitz-Penrose approach but with some assumptions that give some difficulties. This paper proposes a different model that is in accordance with the Lebowitz-Penrose approach, but is able to overcome existing difficulties especially taking into account the fact that for such systems we have many limit cases, special cases, and partial results. The authors propose some tools to describe the metastability phenomenon beyond the case of a dynamics with a recurrence property for a very localized subset of the configuration space -- it is done with the reference to some basic papers. After the presentation of the model (Section 2), the detailed analysis is provided (Section 3) with connections to the exponential law (Section 4) and capacities (Section 5). Section 6 gives soft measures and Section 7 two examples (Glauber dynamics of the Curie-Weiss model and a variation on the so-called ``\(n\)-dog'' theme) that confirm the proposed solutions.