過去十年來，社會科學研究人員對於利用智慧代理人來模擬金融市場市場的接受度已經變得愈來愈高。在這些電腦模擬的虛擬金融市場模型中，就以聖大非研究院（Santa Fe Institute）所提出的虛擬股票市場（Artificial Stock Market）模型最負盛名。因此，有許多學者對此模型提出修正與改進之處，可是卻鮮少為這個模型提出要如何設計或建立模擬系統的細節。雖然，近年來有許多關於這方面的研究，可是他們卻有一些共通的缺點，例如：所設計的系統缺乏可擴充性、便利性，以及友善的使用者介面等等。因此，本研究採用Swarm這個工具以及MASS-Z架構的精神來設計並實作一個多重代理人模擬系統，稱之MASS-S。我們提出MASS-S的架構並將它對映至ASM以說明它如何在MASS-S中運作。此外，我們介紹如何利用Swarm以有效率的方式來實作MASS-S架構。最後，我們利用MASS-S設計不同的模擬實驗來比較其結果。我們相信本研究將有助於不熟悉程式寫作的研究人員能以快速輕易的方式來完成各種模擬系統的建構。 Agent-based simulations of markets have gained more and more acceptance among social scientists in the last decade. The Santa Fe Artificial Stock Market (ASM) is the most well-known one among those computer simulated artificial financial markets. Therefore, there are many researchers to modify the ASM model, but rarely help with the details of designing or building a simulation system for it. There are some efforts supporting the kind of job recently, but they are generally lack of scalability, convenience and friendly user interface of simulation system. Hence, we use Swarm and the idea of MASS-Z to design and implement a multi-agent simulation system, called MASS-S. We propose the architecture of MASS-S and map it to ASM to show how it works. Besides, we also introduce the tool, Swarm, to show how to facilitate it to implement MASS-S. Finally, we design different simulation applications to compare the results of simulations by MASS-S. We believe that it will be a great helpful to researchers to build many kinds of simulation system with an efficient and easy approach, and it will be a contribution to the researchers who aren’t familiar with code writings.
