Vectorizing the fitness and constraints for the genetic algorithm

Hi everyone,

I have an optimization problem with an binary variable, which I'm solving with the genetic algorithm from the Global Optimization Toolbox. I wrote my fitness and constraints with nested for loops. I constrainted the variables to integer and set the lower and upper bounds to 0 and 1. Everything works fine.

Now I wanted to vectorize my code, to improve the speed of the ga. Thats how the my fitness now looks like:

function z = myFitness_vec(y, W, K, M, cst)

cst = reshape(permute(cst,[2 1 3]), 1, numel(cst));
z = bsxfun(@times, cst, y );
z = sum(z);

end

And the constraints:

function [ineq, eq] = myConstraints_vec(y, W, K, M, a, c, t, tc)

yMat = permute(reshape(y, K, M, W), [2 1 3]);

eq1 = sum(yMat(2:11,:,:), 2) - 1;
eq1 = reshape(eq1, 1, numel(eq1));

ineq1_vec = sum(yMat(12:21,:,:), 2) - 1;
ineq1_vec = reshape(ineq1, 1, numel(ineq1));

a = a(:);               
c = reshape(c, 1, []);  
ineq2 = sum(sum(bsxfun(@times, a, yMat), 1), 3) - c;
ineq2 = reshape(ineq2, 1, numel(ineq2));

t = t(:);                   
tc = reshape(tc, 1, []);   
ineq3 = sum(sum(bsxfun(@times, t, yMat), 1), 3) - tc;  
ineq3 = reshape(ineq3, 1, numel(ineq3));

eq = []; 
ineq  = [-eq_1, eq_1, -ineq_1, ineq_2, ineq_3];

end

[With a(1:M), c(1:K), t(1:M), tc(1:K), d(1:K,1:K), i(1:M,1:M), cst(1:M,1:K,1:W) and y(1:M*K*W).]

But now, when I try to use the ga with these functions and the 'UseVectorized' option on 'true' I get an error. If I use the ga without the 'UseVectorized' option it works like with the functions written with loops.

So maybe I just didn't unterstand Vectorization correctly?

[Jan]

Please post the error message and do not let us guess this important detail.