GA’s optimize only combinatorial problems though — where you have discrete set of choices (~genes) for each variable, and therefore do not have a gradient