Feasibility and optimality in model predictive control for linear stochastic systems

By Dr. Luca Bortolussi

We present a technique to provide a stochastic process algebra (SPA) model with a semantics in terms of stochastic hybrid automata. This method has been originally developed for stochastic concurrent constraint programming, but it can be (easily) exported to other SPA, like stochastic pi-calculus or PEPA. The hybrid semantics is actually parametric with respect to the degree of continuity introduced in the model, hence a whole lattice oh hybrid automata is associated to a single SPA system. In this sense, this semantics subsumes both the standard CTMC-based semantics and the more recent, ODE-based, fluid-flow approximation. We will consider some examples and also discuss the asymptotic relationship between different elements of the lattice.