How does penalty factor is applied?

Hello.

I am trying to work with ga() function with integer variables. I want to know how can penalty be applied at unfeasible solutions? How do I define unfeasible solutions as unfeasible?

Actually I am doing this by my own at fitness function. Is it the only way?

Thanks, Vitor Ribeiro.

Answer 1

[Alan Weiss]

It is difficult to know what you are asking. The documentation of mixed-integer ga explains pretty much what goes on in that implementation. If you are doing your own implementation, then you are on your own, unless you give more details and ask a specific question.

Alan Weiss

MATLAB mathematical toolbox documentation

[Vitor Ribeiro]

Hi Alan. I was quiet for a while about this project, but now I am back and I would like you to answer another question regarding GA function on MATLAB. I've read the article about MILP ga, but i don't understand a point.

I am using GA with binary variables and equality constraints adapted by 2 inequality constraints so the sum(x) of the solution vector is equal to Y.

When the algorithm generates new populations applying the special crossover and mutation functions the solutions generated from a feasible one, may now be unfeasible.

So it is not supposed that algorithm evaluate the restrictions and apply penalty function on this unfeasible solutions, rather than evaluate them by fitness fcn?

I have a heavy fitness function and loads of unfeasible solutions that cannot be controlled by input constraints, only inside fitness fcn applying a proper penalty function. So, I would like the algorithm eliminate the ff evaluation of solutions that don't go through basic constraints.

Sorry if my english is a little bit confusing but I hope you can help me out with this.