Parallel Algorithmics of Multi-Agent Systems on MIMD Computers

This book is addressed to Computer Science researchers and professionals who work in parallelism field and/or in multi-agent system (MAS) simulation. They can find in this book interesting analysis of parallel algorithms, but also a simple and powerful model of simulation of multi-agent systems. Agents are autonomous entities acting in an environment, and can be applied to a variety of domains, such as simulation of traffic jams, people or animal populations. The contribution of this book is threefold: (1) provide a simulation model of high-scale MASs, called ParSSAP, (2) do a parallel algorithmic work in MASs, and (3) provide an implementation of this model through a parallel library. The book presents parallelisation issues and the state of the art in MAS simulation. Afterwards, the simulation model and the parallel algorithmics used in two agent percepts (computing of vision fields and potential field propagation) are detailed. Finally, as the library is targeted to easy utilisation and efficient execution, its usage is presented together with a few applications and their execution performance.

Autorentext

Eugen Dedu obtained his PhD thesis in Computer Science in 2002. Since then, he has been writing several papers in international conferences and co-supervised several PhD students. He is currently Assistant Professor in computer networks and teaches in Network and Telecommunications Department of University of Franche-Comté, in Montbéliard, France.