Question

Im trying to implement a GA with tournament selection, but im a bit confused. I know first we select randomly k member from whole population. Then we choose the best one by using tournament size (lets say ts=2). Then we add the winner into the mating pool for crossover. But also I know that I have to fill the mating pool till it reaches the k size again. In this case, if my selected k candidate is always the same selected candidates, wouldn't be the mating pool as same as the selected population? Or Each step I have to select k member for a new winner? a little help pls?

Was it helpful?

Solution

Yes, each step involves selecting a new k elements from the population. Nothing prevents the same element from appearing multiple times and winning multiple tournaments, but unless your tournament size is very large (i.e., approximately equal to the population size), you'll get some randomness in the mating pool.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top