Crossover Variants

[kenje4090]

When we do mutation
1> have list agents for the current population
2> have a list of agents created by mutation, etc (same size as the last list)
3> randomly mix them together into a new list
4> do tournament selection on new list (random pairings) and winner of the tournament (lowest fitness) survives to create a list in 1

add an option for the 4 step cross-over above.

So that the children of winning individuals have to compete against a random individual to survive
“crossover-variant=2"

---

and we could also do “n-random pairings” or tournaments

so we choose two random pairs

• and we do this N times (so 3 rounds) and then we sort individuals by (wins)

so with 3 rounds, we pair that individual with 3 random individuals then we sort individuals by “wins”

and that might be more accurate than a single tournament
then this is option 3
“crossover-variant=3”