著者

川村 秀憲 大内 東 車谷 浩一

出版者

一般社団法人日本機械学会

雑誌

セルオートマトン・シンポジウム講演論文集

巻号頁・発行日

vol.2001, pp.40-44, 2001-11-14

---

The economic and finance systems related to dynamic real world are too huge complex systems to understand the principles of these behavior. Many researchers, i.e., economists, mathematicians, sociologists, computer scientists, and so on, have hoped for long time to model and clarify the inner meaning of such phenomenon. The one of difficulty for these researches is for lack of paradigm that describes the relationship between realistic economic agents with bounded learning or evolving rationality and the emergence macro behavior as aggregation of such agents. Since the first success of research program by Brian Arthur et al., agent-based simulation with bottom-up approach is approved as third mode following theory and experiment against the economic and finance systems. The paradigm of agent-based simulation enables us to naturally introduce bounded learning or evolving rationality to agents. In fact, many researchers study on such systems based on agent-based computer simulation for understanding and elucidating the macro and micro behavior and the emergence between these behavior, and many interesting results are reported based on their computer simulations. It seems for agent-based approach to greatly succeed, however, there are some points at issue in this approach. One point is that these researches generally consist of complicated models and complacently computer simulation programs, and it is unreasonable for other researchers to rubber-stamp the results by a certain researcher. Many models and results are reported, and it is difficult to compare with and refer to these results each other. The another point is that it is difficult to experiment computer simulations for researchers having no skill of computer programming even if they have innovative idea of agent-based model. Especially, this agent-based research program needs participating of many researchers in various research fields for development of this research field, and a researcher well acquainted with computer technology has obligation to provide and develop agent-based simulation tools. According to these background, it is desired to found the forum with common models or simulation tools for promotion of agent-based economics or finance. The X-Economy project aims to found the global standard forum for agent-based economics and finance. As the purpose of first step in X-Economy project, we are concretely struggling to establish the standard model of agent-based economics, the model and rules of self-closed and confrontal game as RoboCup soccer game, the software systems based on open source and communication protocols among such systems, and the software library of various agents in the simulation. The standardization of several ideas of our project, e.g., the softwares, simulation models for researches, results, and so on, is entirely based on open source, and this is the characteristic of our project. Although our project is just beginning, several prototype systems are available, and some research program are worked. In this paper, we introduce the outline of X-Economy system prototype for design and simulation of artificial market. This prototype is based on server-client module structure on TCP/IP connection and XML-based XSS (eXtensible Social System) protocols. This paper includes the image, the components, protocols, implementation techniques of X-Economy tools, and the simulation example and screenshot of our system.