SU Zhao-pin, JIANG Jian-guo, LIANG Chang-yong, et al. An Almost Everywhere Strong Convergence Proof for a Class of Ant Colony Algorithms[J]. Acta Electronica Sinica, 2009, 37(8): 1646-1650.
DOI:
SU Zhao-pin, JIANG Jian-guo, LIANG Chang-yong, et al. An Almost Everywhere Strong Convergence Proof for a Class of Ant Colony Algorithms[J]. Acta Electronica Sinica, 2009, 37(8): 1646-1650.DOI:
An Almost Everywhere Strong Convergence Proof for a Class of Ant Colony Algorithms
Ant Colony Optimization is a novel simulated evolutionary algorithm which has been used successfully to solve many complicated combinatorial optimization problems
but its convergence analysis is seldom researched.The mathematical model of a class of ant colony algorithms is described by TSP problem.On the basis of the decomposition of state space and the construction of reflecting barrier
an almost everywhere strong convergence of the algorithms and the quality that the algorithms can guaranteedly converge to a global optimum set in a finite number of steps are demonstrated by using the martingale theory
and the obtained results may provide a new methodology for convergence analysis of the algorithms.