Tournament selection
I'm trying to implement GA with tournament selection, but I'm a little confused. I know we first choose a random k-term from the entire population. We then choose the best one using the tournament size (say ts = 2). We then add the winner to the crossover twinned pool. But also I know that I have to fill the mating pool until it reaches size k again. In this case, if my selected candidate k is always the same selected candidate, wouldn't the pairing pool be the same as the selected population? Or Do I have to choose a k-member for a new winner at every turn? a little help pls?
source to share
Yes, each step involves the selection of new k elements from the population. Nothing stops the same item from appearing multiple times and wins multiple tournaments, but if your tournament size is not very large (i.e. roughly equal to your population size), you will end up with some randomness in the pairing pool.
source to share