Improved Heuristics for the Minimum Label Spanning Tree Problem

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

“…However, since a higher frequency label may not always be the best place to start, it may not perform as well as the Pilot Method. Another modified version by Xiong et al (2006) is similar to the previous one, except that it tries the most promising 30% of the labels at the initial step. Then it runs MVCA to determine the remaining labels.…”

“…Then, they randomly select one of these candidates, and add it to the incomplete solution. In addition, Xiong et al (2006) presented a Modified Genetic Algorithm (MGA) that was shown to have the best performance for the MLST problem in terms of solution quality and running time.…”

“…Later, a Modified Genetic Algorithm (MGA) was proposed in (Xiong et al, 2006). It outperformed the first two Genetic Algorithms with respect to solution quality and running time.…”

“…Algorithm 3.2: The MGA for the MLST problem (Xiong et al, 2006) Input: A labelled, undirected, connected graph G = (V, E, L) with n vertices, m edges, labels; Output: A spanning tree T ; Initialisation: -Let C ← 0 be the initially empty set of used labels for each iteration; -Let H = (V, E(C)) be the subgraph of G restricted to V and edges with labels in C, where…”

“…Therefore, in the Genetic Algorithm of Xiong et al (2006) the only parameter to tune is the population size.…”