View article

Multi-agent systems (MAS) are often extremely complex and it can be difficult to formally verify their properties. As a result, design and implementation remains largely experimental, and experimental approaches are likely to remain important for the foreseeable future. Simulation is therefore the only viable method to rigorously study their properties and analyze their emergent behavior.

Over the last two decades, a wide range of MAS toolkits and testbeds have been developed and applied in different domains. A number of surveys of these systems have appeared in the literature in recent years, eg,[Serenko and Detlor, 2002; Gilbert and Bankes, 2002; Tobias and Hofmann, 2004; Mangina; Railsback et al., 2006]. However, these have tended to focus on the high level support that these toolkits offer for the process of agent specification and model creation. In contrast this chapter focuses on the simulation engines integrated in to these toolkits to facilitate the execution of the resulting MAS models. In this context a simulation engine is taken to be any virtual machine which provides controlled execution of the model.