Hybrid genetic algorithm for MPRM minimization

• Published in: Zhejiang da xue xue bao. Journal of Zhejiang University. Sciences edition. Li xue ban Vol. 43; no. 2; pp. 184 - 189
• Main Author: Bu, Dengli
• Format: Journal Article
• Language: Chinese
• Published: Zhejiang University Press 01.03.2016
• Subjects: Algorithms; Circuits; Convergence; Genetic algorithms; Minimization; Optimization; Simulated annealing; Strategy; 局部改善; 混合极性reed-muller; 相异度; 逻辑最小化; 遗传算法
• ISSN: 1008-9497
• DOI: 10.3785/j.issn.1008-9497.2016.02.011

Mixed-Polarity Reed-Muller (MPRM) logic minimization is a vital step in the logic synthesis of Reed-Muller (RM) circuits. For MPRM minimization of Boolean functions with large number of inputs, traditional genetic algorithm (GA) is subject to premature convergence. Hybrid GA (HGA) which incorporates local improvement strategy basing on dissimilarity into GA is proposed for MPRM minimization to improve the traditional GA. Local improvement strategy generates a new individual in each iteration by applying crossover operator on the current best individual and the individual which has the maximum dissimilarity to it, then the new individual is used to compete with the best or the worst individual in the population. A set of MCNC benchmark circuits with large number of inputs are minimized by HGA. The results are compared with other intelligent MPRM minimization algorithms. Experimental results show that, the proposed local improvement strategy can help GA escape from the local minima and can enhance global conver