An adaptive genetic algorithm for the minimal switching graph problem

Minimal Switching Graph (MSG) is a graph-theoretic representation of the constrained via minimization problem - a combinatorial optimization problem in integrated circuit design automation. From a computational point of view, the problem is NP-complete. Hence, a genetic algorithm(GA) was proposed to tackle the problem, and the experiments showed that the CA was efficient for solving large-scale via minimization problems. However, it is observed that the CA is. sensitive to the permutation of the genes in the encoding scheme. For an MSG problem, if different permutations of the genes are used the performances of the CA are quite different. In this paper, we present a new CA for MSG problem. Different from the original GA, this new CA has a self-adaptive encoding mechanism that can adapt the permutation of the genes in the encoding scheme to the underlying MSG problem. Experimental results show that this adaptive CA outperforms the original GA.