A One-Parameter Genetic Algorithm for the Minimum Labeling Spanning Tree Problem

The development and application of metaheuristics for problems in graph theory: a computational study

“…Moreover, if C is an optimal solution, then any spanning tree of C is a minimum labelling spanning tree. Thus, in order to solve the MLST problem, it is preferable to seek a feasible solution with the least number of labels (Xiong et al, 2005b).…”

“…For example, Xiong et al (2005b) presented a Genetic Algorithm (GA) to solve the MLST problem, outperforming MVCA in most cases.…”

“…First, those that are reported in the literature to be the best performing are considered, followed by some new approaches. Xiong et al (2005b) presented two slightly different Genetic Algorithms to solve the MLST problem. They both were shown to be simple, fast, and effective.…”

Comparison of Metaheuristics

“…Consider Table 1, which lists basic parameters required for major types of metaheuristics. Though these are guidelines for the minimum For instance, the genetic algorithm for the minimum label spanning tree problem in [32] uses just one parameter, which functions to both control the population size and to serve as a termination criterion.…”

“…In the case of genetic algorithms, for instance, a population size parameter, crossover probability parameter, and mutation probability parameter are typically used, meaning these algorithms will typically have at least the three parameters considered by Deb and Agrawal. However, there have been genetic algorithms developed that operate using only one parameter [32] or none [29], actually eliminating the possibility of parameter interactions.…”