A Design of Problem Solving Environments for Policy Making Assistance Using MAS-Based Social Simulation

In this paper, we propose a design of problem solving environments (PSE) for social scientists. We develop a multi-agent simulation (MAS) for policy making assistance on parallel or distributed computing system such as cluster or computing grid systems, and examine the performance of our approach. Although many applications for PSE have been proposed in natural science such as hydrodynamics and molecular dynamics, there are few applications in social science. We employ an application of MAS-based social simulation to assist policy making in environmental maintenance. Our main purpose in this paper is to show a possibility of social simulation as an application of PSE. We develop a MAS-based application using message passing interface (MPI) and remote procedure call (RPC) on a cluster system. We firstly find the effective condition of MAS-based applications that are suitable for high performance computing systems. Then we examine the their execution time of the application.