The noising method: a new method for combinatorial optimization

This paper presents the principles and the first results of a new combinatorial optimization method that we call the noising method. It is applied to the NP-hard problem of the clique partitioning of a graph. The results obtained from this problem which are definitely better than those provided by standard iterative-improvement methods and compare favorably with those of simulated annealing, show the relevance of the noising method.