A Multi-agent Transgenetic Algorithm for the Bi-objective Spanning Tree Problem

Abstract

The Bi-objective Spanning Tree (BiST) is an NP-hard extension of the Minimum Spanning Tree (MST) problem. The BiST models situations in which two conflicting objectives need to be optimized simultaneously. The BiST has been studied in the literature and several heuristic algorithms were proposed for it, most of them evolutionary techniques. The transgenetic algorithms are among these evolutionary techniques which were successfully applied to the BiST. However, a priori defined parameters can limit the search mechanisms used within the algorithm. In this study, we propose a new transgenetic algorithm for the BiST in which the decision about the search mechanisms used along its execution is automatically made. An analysis of the results of computational experiments carried on 165 benchmark instances showed that the algorithm proposed in this study produces good approximation sets concerning two different quality indicators.