Lab 9 Review by Yoan BOUTE s321389

[YoanBoute]

Hi ! Here's my review on the last version of your work on Lab 9.
Your code has a lot of good ideas, like the Valhalla to always keep the fittest individuals, or the fact that the probabilities of either mutating or recombinating individuals in a certain way evolve with the population. I have some remarks on those ideas :

• You add an individual in the Valhalla each time you generate an offspring. I think it would be way more efficient to first generate all the offspring individuals, and then select in your whole population the 500 fittest individuals to go to the Valhalla, as it would avoid searching through your population 500 times at each generation.
• It seems that the size of your population is not constant, because you start with a 1000 individuals, generating 500 new individuals, and then you kill 1000 of all these individuals, which results in a population of 500. If I understand correctly, you use the individuals from the Valhalla to complete this population, but by doing so you seemingly create a clone for each of these 500 individuals, as they were already the fittest of the whole population. So you still have a 1000 population, but the diversity between individuals is not as high as it should be.
• On the idea of changing the probabilities to pick one genetic operator or another, maybe it could be good to also have a variation in the number of genes affected by mutation, which is here arbitrarily fixed to 5.
  However, I think your code is very interesting and quite well commented, so it's a good work.