Improving this genetic algorithm for error minimization

Question

I\'ve written a simple genetic algorithm, designed to perform fitting. That is, given some input f(x), I\'m able to solve for x without knowing