>1)at crossover, every pair of chromosomes is crossovered? There would be too many chromosomes.
Choose a population size, say N. Given one population of size N, create another one from scratch by repeatedly selecting two chromosomes from the old population, crossing them over and possibly mutating. Algorithm runs faster if you pass the most fit chromosome to the new population unchanged.
>2)how exactly are chromosomes for crossing over choosen?
Straightforwardly: by selecting a number, splitting the chromosomes at this gene, and appending the tails to the wring heads.
Each of the new chromosomes undergoes either a mutation or a gene rotation with probability of 1/2.
(If two original chromosomes had equal fitness value the new chromosomes are rotated with probabilty of 1/2. Otherwise, they mutate with probability of 1/2.)
>3)"On a single evolutionary step either all but the fittest
>chromosome are replaced, or I replace only the two worst
>performers."
>what is the probability of these two cases?
It's rather two different algorithms that are selected with a checkbox.
>4)what is probability that mutation (if is some performed)
>is classic mutation or shift?
I believe #2 above answers that.
>Maybe these questions seem to be too detail, but those
>details are the "thing" which makes algorithm succesful, and
>unfortunately I havent found this settings yet :(.
>I would be very grateful for any suggestions
That's OK, although I had to go through the code I have not used in years.