Theory of λ algorithm

摘要：

　　Genetic algorithm (GA) is most famous and useful algorithm belongs to the class of Evolutionary algorithm (EA).However, often an objective function possesses numerous local optima, which could trap GA from moving toward the desired global solution.GA provides extremely long length of binary string, unsure position of two chromosomes crossover, turbid mutation methods, which fearfully limit the efficiency of GA.In this paper, we propose a new metaheuristic optimization algorithm named as λ algorithm.The new algorithm utilizes strings of digits from member set ”0, 1, 2, 3, and 4” to represent the fitness values of candidate solutions (represented as vectors in n-dimensional Euclidean space).The λ algorithm draws useful information from both repeated and unrepeated digits of strings (candidate solutions), to simulate global advanced schema towards final optimization.The new algorithm only asks 3 or 4 digits to represent an unknown variable,but still could access very precision results.Disciplinary λ comparison and expansion operations instead of inefficient mutation operation, which allowed the strings select more efficient schema from the digits.Without using crossover, stochastic coding strategy, population selection, PSO methods...all of the existing optimization methods, the new algorithm still could achieve the highest searching efficiency, better optimized than most of existing algorithm.

关键词： λ algorithm metaheuristic optimization five elements advanced schema original the highest searching efficiency

作者: Yanghong Cui Renkuan Guo Danni Guo

作者单位: Department of Statistical Sciences,University of Cape Town,Private Bag,Rondebosch,7701,Cape Town,Sou Kirstenbosch Research Center,South African National Biodiversity Institute,Private Bag X7,Claremont

会议类型: 国内会议

会议名称: 第九届中国不确定系统年会、第五届中国智能计算大会、第十三届中国青年信息与管理学者大会

会议地点: 南京

会议语种:英文

页码: 305-320

在线出版日期: 2011-07-27（万方平台首次上网日期，不代表论文的发表时间）