Modeling the Minority-Seeking Behavior in Complex Adaptive Systems

This dissertation, "Modeling the Minority-seeking Behavior in Complex Adaptive Systems" by Fung-kiu, Chow, 鄒鳳嬌, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Abstract of thesis entitled MODELING THE MINORITY-SEEKING BEHAVIOR IN COMPLEX ADAPTIVE SYSTEMS Submitted by Chow Fung Kiu for the degree of Doctor of Philosophy at The University of Hong Kong in December 2003 Minority Game (MG) is a well-known econophysical model which succinctly captures the minority seeking behavior in various economic systems. In MG, a population of heterogeneous selfish players is competing in each turn. In each time step, some current global information is distributed to all players as their basis for choosing one of two alternatives independently; only those who end up on the minority side win. Although its rules are remarkably simple, MG was found to exhibit a surprisingly rich self-organized collective behavior, which is ubiquitously found in many social and economic systems. Indeed, such behavior was also revealed in many statistical mechanical and non-linear physical systems. Thus, MG can be investigated by using the concepts and tools developed by physicists for the study of these physical systems. Unfortunately, some of the assumptions underpinning MG are only valid for idealized economic systems. With this in mind, a number of generalized MGmodels were proposed with the aim of imitating realistic complicated economic systems by simple models. First, two multichoice minority game models known, S) were introduced. as multiple choices minority game (MCMG) and MG(N In these two models, all the strategies are drawn from a set of reasonable, un- biased and computationally feasible strategies. Numerical simulation on these models was carried out and the results were analyzed by non-linear and statis- tical physical tools. From numerical investigation, it was found that these two models exhibit similar global cooperative behavior as the original two-choice MG. Nevertheless, unlike in the MG, periodic dynamics was found to be not always appear in the MG(N, S) model when the strategy space size is much smaller than the number of strategies at play. Furthermore, another generalized MG model called Networked Minority Game (NMG) was proposed. In this game, in addition to global information, certain local information is also used by every selfish player in an unbiased manner for making decision. According to the analysis on the numerical results using physi- cal tools, it was concluded that the level of cooperation in this networked minority game differs remarkably from the original minority game as well as the predic- tion of the crowd-anticrowd theory. The discrepancy from the crowd-anticrowd theory was due to either the broken ergodicity of the local information or the negligence of a four point correlation function in the effective Hamiltonian of the system, depending on the ratio between the two types of information. By studying the above generalized MG models with physical methodology, more and deeper understandings on economy can be achieved. Such knowl- edge may contribute to the search of the fundamental principles governing the economic systems, which have not been all revealed by conventional economic theory. DOI: 10.5353/th_b2936748 Subjects: Economics - Mathematical models Economics - Statistical methods Statistical physics