Defining Objective Function in Genetic Algorithm

Trying to optimize the following function using Genetic Algorithm (A,B are arrays of length 10000 attached)

y= [x] * [A-B]

such that

• x is binary
• -B'*x <= blimit - sum(B)

load ('A_B_Arrays');

nvars = numel(A);
blimit = 1165.208;

Aineq = -B';
bineq = blimit - sum(B);
Aeq = [];
beq = [];
lb = zeros(1,N);
ub = ones(1,N);
nonlcon=[];
intcon = 1:nvars;

tic
x = ga(fun,nvars,Aineq,bineq,Aeq,beq,lb,ub,nonlcon,IntCon)
toc

MATLAB Documentation says that objective function should accept "a row vector of length nvars and return a scalar value". The length of vector x, in this case, depends on the length of [A-B]. How do I define the objective function with so many input variables?

function y = fun(x(1),x(2),x(3),...,x(nvars))
y=[x(1) x(2) ... x(nvars)] * [A-B];
end

Answer 1

[Walter Roberson]

Your M is undefined in optfun. You should see http://www.mathworks.com/help/matlab/math/parameterizing-functions.html

[Neeraj Rama]

Thanks!

Answer 2

[Geoff Hayes]

Neeraj - your fitness function would be simply

 function y = fun(x)
 y = x * [A-B];
 end

where x is an array of variables being optimized.

[Neeraj Rama]

Okay, I understand. But the following error pops up, "Failure in user-supplied fitness function evaluation. GA cannot continue". Rechecked the definitions of input arguments but couldn't trace the issue.

load ('A_B_Arrays');

M=A-B;

nvars = numel(A);
blimit = 1165.208;

fun=@optfun;
Aineq = -B';
bineq = blimit - sum(B);
Aeq = [];
beq = [];
lb = zeros(nvars,1);
ub = ones(nvars,1);
nonlcon=[];
IntCon = 1:nvars;

tic
x = ga(fun,nvars,Aineq,bineq,Aeq,beq,lb,ub,nonlcon,IntCon)
toc

   function y = optfun(x)
   y = x * M;
   end