Research On Distributed Multi-Agent Simulation Technology&Application

Complexity Theory is a pilot research field in modern science and it’s the main task inthe research of complex system. Complexity science is treated as the science of21thcentury. The research purpose is to discover the system dynamics which can not beexplained by current methodology.Multi Agent Simulation is a research approach for complexity science which is basedon the Multi Agent System theory and Modeling&Simulation theory. Distributedprocessing is a hot issue in the research and applications of Multi Agent Simulation. Thispaper presents a design for Generic Distributed Multi Agent Simulator while it is still in anexperimental phase of this research area. The targets of this simulator are the highparallelism, generic utilization and distributed processing.Multi Agent System (MAS) as a new theory and approach for applicationdevelopment has been bought to attention, while its practical utilization is limited by thetechnical provision and methodology. The multi agent programming language is afoundation for developing multi agent applications and carrying out multi agentsimulations. The current researches on multi agent modeling and programming languageare focused on the semantics, Integrated Development Environment (IDE), agentframework and the underlying multi agent platform more or less. This paper will raise thequestions in designing a MAS programming language and present a new multi agentmodeling language named MA-UML which is modeling paradigms independent.We introduce the multi agent simulation theory and several multi agent simulationenvironments as well as the difficulties that researchers are facing while carrying outsimulation experiments. The latter of this paper discussed the problems to be solved inthe research of generic multi agent simulator and present an initial design of it.Financial and Economic simulations are popular applications in complex systemsimulation. The main task includes platform archetecture, database, modeling of behaviorand data analysis. We developed a prototype based on a framework we initiated which canbe applied in financial and simulation data analysis.