IMPLEMENTATION AND EVALUATION OF GENETIC ALGORITHMS FOR SYSTEM PARTITIONING

The implementation and evaluation of several algorithms to solve the partitioning problem are described. Partitioning is performed according to the main strategy of, at each step, separating the elements into two groups and seeking to minimize the connections between elements belonging to different groups. Several algorithms are considered; in particular, the main aim of this work is to evaluate algorithms and to consider the possibility of improving them by introducing steps based on the use of 'hill climbing' and simulated annealing algorithms. The numerical results of many computer simulations are presented to outline the main features of each algorithm and to give an evaluation of the most suitable way of solving the partitioning problem.