Hello all
I have a problem with the Niched Pareto GA of Horn et al. (1993) in
solving a multi-objective problem., i.e it doesn't converge to the set
of true
non-dominated solutions.
Let me describe briefly how do I select individuals in the NPGA:
Two individuals are picked at random from the current generation along
with a comparison set (tournament). The two individuals are compared
against the individuals of the tournament. If one is dominated and the
other is not, the latter is selected. If both or none of them are
dominated, the one with the smaller niche count is selected. This
procedure is continued untill the intermidiate generation is filled up.
Then reproduction, crossover and mutation are applied to form the next
generation using the individuals in the intermidiate generation.
Does anybody think something is missing in this algorithm?
Any help or direction would be greatly appreciated.
Afshin
--
Seyed Afshin Mansouri
Ph.D Student,
Department of Industrial Engineering,
Amirkabir University of Technology,
Tehran 15914, Iran
E-Mail: [log in to unmask]
Fax: +98 (0)21 6413025
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|