why ga can not disperse population in a feasible area ??

I employed ga to minimize a single objective problem with 56 variables. My constraints include linear equalities and inequalities. There are two linear equalities and 52 inequalities. I try to confine variables by defining linear inequalities. I should mention that variables are dependent, so these linear inequalities are linear but they are related together. I run ga through optimization tool but the minimum value is nearly the same in all iterations of each run. I think the problem is the population generation because in each iteration I can not see even a small change in variable distribution, the variables are always constant. I try to change options for the population part, but I didn't see any improvement. I checked the code and I'm totally sure that the problem should be somewhere in ga options.

[Rik]

Have a read here (or here for more general advice) and here. It will greatly improve your chances of getting an answer.

Answer 1

[Alan Weiss]

The real question is why are you using ga to solve this problem? If your objective function is smooth, use fmincon starting from a variety of initial points; do this automatically using MultiStart. If your objective function is not smooth, use patternsearch as suggested here.

Alan Weiss

MATLAB mathematical toolbox documentation

[abouzar jafari]

Hi Alan I employed fmincon and I got good results. tnx a lot.