No abstract

No abstract

International audienceIn large scale agent-based simulations, memory and computational power requirements can increase dramatically because of high numbers of agents and interactions. To be able to simulate millions of agents, distributing the simulator on a computer network is promising, but raises some issues like: agents allocation and load-balancing between machines. In this paper, we study the best ways to automatically balance the loads between machines in large scale situations. We study the performance of two different applications with two different distribution approaches, and we show in our experimental results that some applications can automatically adapt the loads between machines and get alone a high performance in large scale simulations with one distribution approach than the other

Abstract. In the field of multi-agent based simulation (MABS), the concept of environment is omnipresent, though poorly defined. We argue here that depending on the modeling of space and of relations between agents, only a few efficient implementations can be set up. We aim at formalizing the core functions of environments, so as to highlight the computational answers to possible modeling choices. This unifying approach leads to the identification of four paradigmatic Design Patterns, associated with specific atomic environments, which can be composed in order to tackle complex situations.

In the context of situated agents simulations, when the number of agents increases, the number of their interactions will be increased too. These growths leads to higher requirements in memory and computation power. When simulations involve millions of agents, it becomes necessary to distribute the simulator on a computer network. In this paper we study the impact of synchronization policies in such context. Our claim is that when millions of agents are used in a simulation, because observations of these complex systems is made at the population level, emergent properties at the macroscopic level should not be highly impacted if some failure appears at the microscopic level. This paper is focused on the study of the impact of synchronization relaxation in the context of large scale situated agents simulations. We evaluate the cost in performance of several synchronization policies and their impact on the macroscopic properties of simulations. To that aims, we study three different time management mechanisms and evaluate them on two multi-agent applications.