Re: Question about wich elements are taken to the new generation

none wrote:
Let's say we have the following initial population

0001 = P0
1001 = P1
1010 = P2
1110 = P3

We rank the four with a fitness function

P0 = 19
P1 = 10
P2 = 15
P3 = 18

Now we chose the best for reproduction.

Let's say roullet picks P0 and P2. We cross them and get two new
elements for the population.

Previous generation: P0 P1 P2 P3
New elements: P4 P5

* Do we keep crossing until we have a new generation of N elements (in
this case four) making that all members in the next generation are
completly new
or
* Do we take some of the elements in the current generation to the next one

My understanding is that you can do all sorts of things, including both
of what you've selected. For instance, based on your example (if higher
== more fit), you could take P0 and P3 as the 2 most fit chromosomes
and insert them into the new population without crossover or mutation.
Then you would only need to generate 2 children based on your roulette
wheel. I believe that is referred to as elitism. I've also seen
algorithms that "grow" the population during iterations, so you do not
have to always completely replace the previous generation. It seems
that it all depends on how you want to configure your environment.

.