A two-phase local search for the K clusters with fixed cardinality problem

In this paper we present a two-phase local search procedure to obtain a feasible solution for the K clusters with fixed cardinality problem. In order to evaluate the performance of this local search procedure for this NP-hard problem, a computational experiment is designed to compare the cost of the computed feasible solution either with the optimum value, or with the upper bounds obtained from the linear relaxation of a strengthened mixed-integer linear formulation defined for this problem, by using a standard software.