An analysis of MML agent simulation under uniform distribution of gain parameter

Koichiro Ishikawa, Bongsung Chu, Akito Sakurai, Hiroaki Matsukawa

Research output: Contribution to journalArticle

Abstract

We often use uniform distribution to describe diversity of agent values. And this description often used in agent simulation which mimics real society on computer which we call an artificial society. In this paper we show that, an agent simulation under uniform distribution may cause wide divergence of final convergent consensus points, though we give a fixed initial macro consensus value to an artificial society facing an alternative decision. Because intuitive observation provides a same convergent value of final consensus for a given initial value of macro consensus, we call this uncertainty or divergence of the final consensus the pitfall of agent simulation. Through numerous simulations, we found that the divergent final consensus have some order and the mechanism of the uncertainty were unveiled in this paper.

Original languageEnglish
Pages (from-to)4693-4703
Number of pages11
JournalInformation (Japan)
Volume16
Issue number7 A
Publication statusPublished - 2013 Jul

    Fingerprint

Keywords

  • Agent simulation
  • Micor-macro loop
  • Uncertainty

ASJC Scopus subject areas

  • Information Systems

Cite this