摘要
鉴于羊群行为的复杂性和重要性,基于消费者行为的最新研究成果,以多智能体建模的研究方法仿真并对比分析不同类型复杂网络对羊群效应现象的影响。模型中,决策者的羊群行为采用最新的实证研究成果。通过仿真,不仅验证了数学建模分析能够得到的结论,还因过程仿真而能够发现羊群效应现象的时间特性。通过对常见的小世界网络和无标度网络环境下的羊群行为仿真对比,发现网络结构是决定羊群效应和厚尾效应显著性的关键因素之一。例如,"棋盘式"网格网络和E-R随机网络环境下的羊群效应更加明显,小世界网络环境下的羊群效应相比之下并不明显,而无标度网络环境会出现厚尾现象等。另外,仿真结果表明复杂的消费者行为与复杂的网络结构叠加后,其对羊群效应现象的影响反而并不显著。
Herd behavior is valuable but complicated.The impact of different type complex network on herd effect is analyzed by multi-agent modeling and simulation.In the model,herd behavior of decision-maker is modeled based on the advanced evidences from empirical study.Through simulation,not only the conclusions drawn from traditional mathematical modeling are verified,but also the whole evolving process of herd effect is inspected and the evolving speed related features can be used to evaluate performances of herd effect.More important,simulation results show the network structure of different type complex networks is one of the key factors for herd effect and heavy-tail effect.For instance,homogenous networks,i.e.regular grid network and E-R random network present a significant herd effect,small world network shows less herd effect,while heavy-tail effect can be found in scale free network environment.When combining with heterogeneous random network,the influence of non-linear herd behavior is not obvious for herd effect.
作者
李锋
魏莹
Li Feng;Wei Ying(School of Business Administration,South China University of Technology,Guangzhou 510640,China;Department of Business Administration,Jinan University,Guangzhou 510632,China)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2021年第3期539-553,共15页
Journal of System Simulation
基金
国家自然科学基金(71572070)
广东省哲学社会科学规划项目(GD20CGL20)。
关键词
羊群效应
消费者行为
复杂网络
多智能体建模与仿真
herd effect
consumer behavior
complex network
multi-agent based modeling and simulation
