Sean,

thanks for your answer. I explained my work in this old topic: https://listserv.gmu.edu/cgi-bin/wa?A2=ind1312&L=ECJ-INTEREST-L&P=R2&X=6E553B5D8D0C2B5870&Y=victorjatoba10%40gmail.com
But I changed my direction and implemented it using GA and not GP.
So I really need of one GeneVectorIndividual cause my individual should be a vector of Genes. But you're right, I have 4 ways to mutate my gene: 

1. Inserting a discipline and their workload in a random period of the a also random day
2. Removing a discipline...
3. change the discipline...
4. Change the workload.

The problem is in the second way, when removing a discipline. So I remove this way and my populate increase in the next generations having a nice behavior.

So, I stil have a question. Two individual are selected for crossover and mutation process, but is generate two copies and not the real individuals, after the processes spoke before, these two clones are added in the new Subpopulation and the two parents are mantained in the Subpopulation too (right?). If I have many Subpopulations (> 1000), in the beginner generations, the algorithm is very fast but after to more generations the algorithm is not too fast as before. Why does it occur? It seems that the subpopulations is accumulating the number of individuals. How the algorithm define the fitness value to kill the weakest individuals? 

Thanks for all.

Regards.


On Thu, Feb 20, 2014 at 10:06 AM, Sean Luke <[log in to unmask]> wrote:
Did you override the Gene.mutate method to mutate in some way appropriate to your problem?  By default all it does is randomize the gene (it calls reset(), which may be too extreme). Without any mutation, you're relying solely on crossover.

Are you sure you need GeneVectorIndividual?  You can't do your task with IntegerVectorIndividual, say?

Sean

On Feb 20, 2014, at 12:20 AM, Victor Miranda Gonçalves Jatobá <[log in to unmask]> wrote:

> Hello,
>
> I'm working with a GA problem. For this I implemented the Gene and the
> GeneVectorIndividual. I also had to override the crossover and mutation
> methods. The weird of the algorithm behave is that if I set the mutation-prob
> value different of zero my population don't evolve and the best individuals
> present on the final chart pdf in each generation are very oscillate (see
> Mutation.pdf attached). In the other way, if I put the mutation-prob to zero,
> the population are evolve gradually, but this evolution is little and more stable,
> because don't occurs mutation (see Crossover-Any.pdf attached). Anyone
> know what is occurs and how do I fix it?
> It seems that after mutate process, the parents (fittest of the population) of
> the clones are not put in the new Subpopulation then next generations never
> evoluate.
>
> My main params:
>
> pop.subpop.0.species.crossover-prob = 0.25
> pop.subpop.0.species.crossover-type = any
> base.likelihood = 0.5
> pop.subpop.0.species.mutation-prob = 0.1
> pop.subpop.0.species.pipe = ec.vector.breed.VectorMutationPipeline
> pop.subpop.0.species.pipe.source.0 = ec.vector.breed.VectorCrossoverPipeline
> pop.subpop.0.species.pipe.source.0.source.0 = ec.select.TournamentSelection
> pop.subpop.0.species.pipe.source.0.source.1 = ec.select.TournamentSelection
>
> select.tournament.size = 2
>
> Best Regards,
> Victor jatobá

>





--
att,
Victor Jatobá

“A imaginação é mais importante que o conhecimento. O conhecimento é limitado. A imaginação envolve o mundo.” Albert Einstein.