Non-expansive ε-bisimulations for probabilistic processes

  • Authors:
  • Simone Tini

  • Affiliations:
  • Dipartimento di Informatica e Comunicazione, Università dellInsubria, 22100 Como, Italy

  • Venue:
  • Theoretical Computer Science
  • Year:
  • 2010

Quantified Score

Hi-index 5.23

Visualization

Abstract

@e-bisimulation equivalence has been proposed in the literature as a technique to study the concept of behavioral distance between probabilistic processes. In this paper we first consider the generative model of probabilistic processes and introduce two stronger equivalence notions: action@e-bisimulation and global@e-bisimulation. For each of these three equivalence notions we propose an SOS transition rule format ensuring the property of non-expansiveness. Non-expansiveness means that if the behavioral distance between s"i and t"i is @e"i, then the behavioral distance between f(s"1,...,s"n) and f(t"1,...,t"n) is no more that @e"1+...+@e"n. As expected, the stronger the @e-bisimulation considered, the weaker the constraints of the transition rule format. Then, we switch to the reactive model of probabilistic processes and we propose a rule format for @e-bisimulation and action@e-bisimulation, arguing that global @e-bisimulation is not needed in such a context.