Multicriteria Partitioning Problems

In this paper we study the possibility of generalizing a monocriteria model to a multicriteria partition problem and show the reason why exact algorithms are not efficient in solving the problem. The mathematical model studied is a modified version of a mono-objective model from Mehrotra [1] and applied to the biobjetive case using the ε-constraint method. Research conducted concludes that the studied multicriteria model is not suitable to solve the problem due not only to the high computational requirements needed to run the model, as well as the difficulty of ensuring the fundamental constraints of integrality, contiguity and ―absence of holes'''. Kruskal's algorithm applied to multicriteria partitioning problem [2] shows satisfactory results, managing to secure the general constraints of contiguity and integrality without requiring a high computational effort, with very low execution times compared with the exact method.