Abstract
This paper describes a method to design a large-scale social simulation with the environmental information on parallel computer. To implement a large-scale Multi-Agent Simulation (MAS), a large number of agents and a large size of environmental information should be implemented in the simulation program. In this paper, we propose a representation technique to implement a simulation program on parallel computer. Our proposed technique is to divide environmental information into some sub-environmental information. We firstly explain the proposed technique and a rule which mediate between agents for parallel MAS. We illustrate how to implement sugarscape-based simulation program with the proposed technique for parallel computer system. Then we evaluate the simulation program on a PC cluster system for the large-size environmental information. The simulation results show that there is no significant gap between the experimental data obtained by the proposed technique and the non-parallel technique. We also show that the computation time of the simulation program with the proposed technique can be improved on a PC cluster system.
