Genetic algorithm for linear and cyclic assignment problem

A hybrid algorithm based on a genetic approach and a pairwise exchange procedure is developed for solving the linear and cyclic assignment problems. A technique which minimizes the amount of calculations during the exchange procedure is used. A comparison is made between the quality of solutions obtained by the genetic algorithm, and solutions achieved by another efficient algorithm based on a constructive procedure. Evaluation of modifications to the basic genetic procedure assists in recommending an improved algorithm. These modifications include various reproduction and selection rules, and changes to initial population size.