genetic algorithm c++ final stage

Question

I have written the code for a genetic algorithm and have the crossover function, mutation function, parent selector function and a function to transfer genes. Now I want to put it all together and would like to know if the following code is good programming practice or not.

Species Parents[popSize];
Species Children[popSize];

for(int gen = 0 ; gen < 100 ; gen++)
{
    for(int i = 0; i < popSize ; i ++)
    {
        int parentA = chooseParent(Parents);
        int parentB = chooseParent(Parents);
        crossOver(Parents[parentA] , Parents[parentB] , Children[i]);
        Children[i].mutate();
    }
    for(int i = 0; i < popSize ; i ++)
    {
        transfereGenes(Children[i], Parents[i]);
    }
}

I think this should be on https://codereview.stackexchange.com/ — Thijser, Oct 26 '15 at 20:57
Im sorry i was unaware. Is it in violation of this stackexchange? — user5351703, Oct 26 '15 at 20:59
Well stackoverflow is mostly for questions of "how do I do X" while codereview is more about "I wrote this piece of code what do you think?" So if you move it you are far more likely to high quality comments. — Thijser, Oct 26 '15 at 21:01

score 0 · Answer 1 · edited May 23 '17 at 12:25

It's a simple, clear, standard implementation of a genetic algorithm. It's enough for many application.

Anyway you could consider some points:

using std::vector instead of arrays
1. it's a way to gain flexibility without losing speed (e.g. there're some variations of the basic genetic algorithm that consider variable-size populations)
2. it allows to swap Children and Parents via the swap function (faster for large populations)
you can probably change the chooseParent function to extract a couple of individuals (and, in general, n individuals acting as parents) in a single "step". This is quite simple for tournament selection.
you should add some control parameters (probability of mutation, crossover...). Many problems are very sensitive to these parameters and you can consider different values during evolution.
for strictly C++ related points:
1. you should avoid the postfix increment operator (gen++, i++)
2. 100 is a magic number

genetic algorithm c++ final stage

1 Answers1