Using parallel computing to improve the scalability of models with BDI agents

Patrick Taillandier (1) , Mathieu Bourgais (2, 3) , Alexis Drogoul (4) , Laurent Vercouter (5)

1 MIAT INRA - Unité de Mathématiques et Informatique Appliquées de Toulouse
2 IDEES - Identité et Différenciation de l’Espace, de l’Environnement et des Sociétés
3 LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
4 UMMISCO - Unité de modélisation mathématique et informatique des systèmes complexes [Bondy]
5 MIND - LITIS - Equipe Multi-agent, Interaction, Décision

These last years have seen the development of several extensions of modeling platforms to include BDI agents. These extensions have allowed modelers with little knowledge in programming and artificial intelligence to develop their own cognitive agents. However, especially in large-scale simulations, the problem of the computational time required by such complex agents is still an open issue. In order to address this difficulty , we propose a parallel version of the BDI architecture integrated into the GAMA platform. We show through several case studies that this new parallel architecture is much more efficient in terms of execution time, while remaining easy to use even by non-computer scientists.

Mots clés: Parallel Computation; Cognitive Agent; Modeling Platform; Simulation

Domaines: Modélisation et simulation

Patrick Taillandier, Mathieu Bourgais, Alexis Drogoul, Laurent Vercouter. Using parallel computing to improve the scalability of models with BDI agents. Social Simulation Conference, Sep 2017, Dublin, Ireland. ⟨hal-01573385⟩