Basic question about Selection Strategy and Mutation Operator

[Vikas Agrawal]

I have been reading about this for a while, but still confused to a certain extent.

 

 

When it comes to basic genetic algorithm process, which one of the following two is correct?

 

 

1) First generate initial candidate solutions for first generation -> then perform selection strategy -> then create a pool of selected candidate (how big if so?) -> then select any two from such pool and perform mutation and crossover function -> and finally put them in the next generation. Then again select any two from the pool of selected candidate and perform the mutation and crossover function -> put them in the next generation. Repeat the process until the population size is reached.

 

So when the Selection Strategy is performed, one complete pool of selected candidates is created. I am assuming that this pool size is defaulted to the population size. Once such pool is created, mutation and crossover operations are performed by randomly selecting two elements.

 

or 

 

2) Generate candidate solutions -> perform selection strategy -> directly perform mutation and crossover function on such two selected candidate (i.e. no pool is created) -> put mutated candidates into next generation. Then again perform the selection strategy process  -> perform mutation and crossover function -> put mutated candidates into next generation. Repeat the process until the population size is reached.

 

So in this case, Selection strategy is immediately followed by Mutation/Crossover operation. So for every new candidate in the next generation, first Selection Strategy is performed and then the mutation/crossover.

 

I hope I don't sound too confusing. 

 

Thanks,

 

Vikas

 

×Proofread

[klaas hoelscher]

If i remember correct, it works this way:

1. Generate Population and..

2. evaluate fitness for all individuals

3. set aside a copy of x from 0..n members of said population if eligible for elite selection

4. create new Population of size n-x with copying members as selected by your chosen selection algorithm

5. mutate/cross individuals of new Population

6. copy elite members from step 3. your population is now at n members again

7. start over at 2.

this works out to option 2) of yours. Meaning: unfit members don't get the chance to reproduce

memory may be foggy. you could always look it up in the code, its pretty easy to understand

regards,

Klaas

--
You received this message because you are subscribed to the Google Groups "Watchmaker Framework for Evolutionary Computation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to watchmaker+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Vikas Agrawal]

That makes sense. Many thanks for explanation.