Is it possible to use a matrix as a decision variable for GA?

For my optimization problem, I have a 13x144 matrix as a binary decision variable. Since it is integer and non-linearly constrained, I cannot use fmincon. Is it possible to use GA? If so, how do I define nvars, lb, and ub?

If ga will not work, is there another Matlab solver that I should use for this kind of problem?

I'm running Matlab R2014a.

Thank you

Answer 1

[Alan Weiss]

All you have to do is take your matrix of decision variables and reshape it to a row vector for use in GA. For example, if x0 is a 13x144 matrix, then

 b = (x0(:))';

is a row vector. To turn the row vector back into a matrix, use reshape:

 c = reshape(b,size(x0));

That said, I do not think that you will have much luck with that many decision variables and GA. I believe GA is suitable for perhaps up to 100 decision variables, not thousands. You would probably have better luck using intlinprog with a linear approximation to your constraints, along the lines of this example.

Good luck,

Alan Weiss

MATLAB mathematical toolbox documentation