Application of the H-method to solving combinatorial optimization problems on permutations

Abstract

A hybrid algorithm class for combinatorial optimization problems solving is considered. The class is named the H-method and built on the basis of synthesis of the accelerated probabilistic simulation algorithm (G-algorithm) and the modified discrete downhill simplex method. Algorithms for segment and half-interval building in the permutation space are proposed and proved. They are used for quadratic assignment problem solving. The computational experimental results are given, which demonstrate a high efficiency of the developed algorithms in comparison with some known algorithms.