Peew review for lab2

[giogenna16]

Hi, Overall, I think that your results are good and competitive; the code is well written and, thanks to the explanation on PDF and the comments, it is easy to understand it. As emerged during lectures, it is foundamental for this problem to take care of the proportion between zeros and ones in the gemomes which constitute the population: it is possible to reach, on average, better results maily in two ways:

1. setting more zeros than ones when the population is created (so in the initialization), as you decided to do;
2. creating mutations which tend to set up gemomes with more zeros than ones (you did something similar at the end of the algoritm).

I noticed that, in the fitness, if you change the number of repetitions in the list with the length of the list, the results are the same, because the logic is the same, but there is a minimum reduction of the computational cost. It is interesting the intuition, that brings to better solutions, to combine a pure GA with a local search focused only on the mutation of the best indviduals, so doing exploitation when the search enters in a steady state. Another approach that you could try to implement to improve the solutions is using a pure GA, but, creating different mutatios (for example, as already mentioned, yours and another which tends to set up gemomes with more zeros than ones) and, on the base of the number of generations done so far, change the probabilities to do a certain mutation with respect to the other and the crossover. This could help also to reduce the execution time, maybe mantaining results similar to that you obtained.

[FlavioPatti]

Thank you very much for the review!