A First Approach of Grouping Problem in Stochastic Automata Network - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Rapport Année : 2004

A First Approach of Grouping Problem in Stochastic Automata Network

Dominique Barth
  • Fonction : Auteur
Johanne Cohen
Mathieu Le Coz
  • Fonction : Auteur
Franck Quessette

Résumé

In the Stochastic Automata Network (SAN) formalism, a SAN is composed of a set of automata and a set of synchronizations which model an underlying Markov chain. For each automaton and each pair of automaton-synchronization one has to store a matrix. This leads to a low storage compared to the storage of the underlying Markov chain but the time complexity depends on the number of automata and synchronizations. We propose to contract automata together in order to get a contraction of the number of automata and synchronizations and a reduction of time complexity at the price of a larger storage. In this first work, we assume no function, we do not take into account, the reachability function and we assume dense storage for the matrices. Assuming a maximal amount of storage capacity, we prove that to find the best contraction, under our assumptions, is a NP-complete problem. Then we give a heuristic and make some comparisons between the heuristic and the best value and we show that this heuristic gives quite good results.
Fichier non déposé

Dates et versions

inria-00100238 , version 1 (26-09-2006)

Identifiants

  • HAL Id : inria-00100238 , version 1

Citer

Dominique Barth, Johanne Cohen, Mathieu Le Coz, Franck Quessette. A First Approach of Grouping Problem in Stochastic Automata Network. [Intern report] A04-R-072 || barth04e, 2004. ⟨inria-00100238⟩
67 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More