Evolutionary operators for the Hamiltonian completion problem

Abstract

This paper deals with evolutionary algorithms for solving the Hamiltonian completion problem. More precisely, the paper is concerned with a collection of crossover and mutation operators, which mostly originate from the traveling salesman problem, but have further on been modified or customized for Hamiltonian completion. The considered crossovers and mutations are tested on a set of randomly generated problem instances. The obtained experimental results clearly show that the behavior and relative ranking of the operators within the Hamiltonian completion environment are different than within the traveling salesman environment. Moreover, it is shown that our modified or custom-designed operator variants accomplish much better results for Hamiltonian completion than the standard variants.