Learning Evolutionary Complex Systems through Designing a Computer Simulation

• Main Authors: Ying-Hong Liao, 廖英宏
• Other Authors: Chuen-Tsai Sun
• Format: Others
• Language: en_US
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/91239437185113888131

博士 === 國立交通大學 === 資訊科學系 === 90 === Interest in complex systems (CS) has been rapidly growing over the last decades. The field has rapidly matured as evidenced by the increasing number of conferences, workshops, and papers as well as the emergence of a central journal for the field. CS is highly promising for many biological, ecological, engineering, economical, and sociological applications owing to their generality and robustness. Among numerous CSs, the dynamically evolving evolutionary complex systems (ECS) are more applicable to real world problems, such as ecological systems, dynamic equilibrium economical systems, and social modeling systems. An increasing number of science and engineering schools are thus offering related courses. Students enrolled in a ECS-related course not only study the evolutionary process of ECS, but also learn how to manipulate diverse range of ECS techniques. Our experience of investigating and teaching ECS-related courses has revealed the essential role of experimental design and simulation in practically implementing ECS. Therefore, this work presents a learning environment with three simulation systems to assist students in implementing ECS, with particular emphasis on the vital role that practical experience plays in computer-related courses. In this work, we first identify important issues and challenges encountered in ECS learning, especially in the domain of evolutionary computation (EC). We then summarize important factors mentioned in previous constructive epistemology and propose pedagogy goal for learning: learning through designing a computer simulation. After setting the goals we introduce a framework and guidelines of an instruction strategy that can explore the niches provided by simulation-based learning environments. The proposed strategy is learning through designing a computer simulation that is a way of learning by doing. We emphasize simulation modeling and creativity during the learning process. Three simulation systems are then proposed and adopted to assist learners in different levels: beginning learners, intermediate learners, and advanced learners. These programs relieve students of detailed programming so they can concentrate on grasping principles of ECS. These programs also outlines how to design evolutionary simulations and how to set parameters of different kinds of problems. The purpose of making these systems available is to encourage the experimental use of ECS on realistic problems, and thereby to identify the strengths and weaknesses of ECS. This work describes the programs, shows how they can be used to solve various problems, and provides details on their implementations. Also described herein is our experience of using these software to assist students in designing and simulating experiments. Experimental results demonstrate the merits of adopting computer-based experimental design and simulation in learning ECS.