Study on Probability Distributions for Evolution in Modified Extremal Optimization
Guo-Qiang Zeng,Yong-Zai Lu,Wei-Jie Mao,Jian Chu
DOI: https://doi.org/10.1016/j.physa.2009.12.055
IF: 3.778
2010-01-01
Physica A Statistical Mechanics and its Applications
Abstract:It is widely believed that the power-law is a proper probability distribution being effectively applied for evolution in tau-EO (extremal optimization), a general-purpose stochastic local-search approach inspired by self-organized criticality, and its applications in some NP-hard problems, e.g., graph partitioning, graph coloring, spill glass, etc. In this study, we discover that the exponential distributions or hybrid ones (e.g., power-laws with exponential cutoff) being popularly used in the research of network sciences may replace the original power-laws in a modified tau-EO method called self-organized algorithm (SOA), and provide better performances than other statistical physics oriented methods, such as simulated annealing, tau-EO and SOA etc., from the experimental results on random Euclidean traveling salesman problems (TSP) and non-uniform instances. From the perspective of optimization, our results appear to demonstrate that the power-law is not the only proper probability distribution for evolution in EO-similar methods at least for TSP, the exponential and hybrid distributions may be other choices. (C) 2010 Elsevier B.V. All rights reserved.