Patter search optimization, not getting fittest solution

Have adapted a code a GA code to run with PS but it get to a solution after a few iterations and can not find a fitter solution. I can manually find better solution by changing around the variable.

Keeps on refining the mesh without finding a better solution:

48 393 36176.4 9.537e-07 Refine Mesh Optimization terminated: mesh size less than options.TolMesh.

Answer 1

[Alan Weiss]

I suggest that you put finite bounds on each variable, lb and ub. Then start your optimization at several random initial points:

 x0 = lb + rand(size(lb)).*(ub - lb);

Alan Weiss

MATLAB mathematical toolbox documentation

[Mo]

How do I give it several initial points? There doesn't seam to be way of defining the number of variable to optimize, so it takes the entire size of the array of initial points as the number of variables.

[Alan Weiss]

What I mean is, give points one at a time as you run the optimization multiple times.

Alan Weiss

MATLAB mathematical toolbox documentation