Error "Your fitness function must return a scalar value"

Hello everyone,

I'm trying to use ga algorithm optimization with non linear inequality constraints by applying penalty method.

But I have the error :

       ??? Error using ==> makeState at 56
       Your fitness function must return a scalar value.

       Error in ==> gaunc at 41
       state = makeState    
       (GenomeLength,FitnessFcn,Iterate,output.problemtype,options);

       Error in ==> ga at 279
       [x,fval,exitFlag,output,population,scores] = gaunc(FitnessFcn,nvars, ...

       Error in ==> ga_test at 55
       [x_opt,fval_opt,exitflag_opt] = ga(ObjectiveFunction,nvars,[],[],[],[],
       [],[],[],options)

for the following reason:

In fact, i need to apply a constraint which depends on values coming from excel table. This excel table contains 2 columns:

       rows     x1     y1

         1      4.5    10
         2      5.5    11
         3      6.5    13
etc.

The fitness function has 6 variables. 2 of them (var1 and var2) have to respect the following condition:

        var1/var2 = ratio : i have to find which value of x corresponds 
        to the ratio value,

        so if var1/var2 = 5.6, so y1(find(x1 <= var1/var2, 1)) = y1(2) = 11
        and this 11 value will be use into an equation which has to be        
        positive as a constrained condition.

if I'm not clear enough, example of constraint :

        constr = ((y1(find(x1>=L./(x(2)./2), 1 ))).*E.*(x(1)./(x(2)./2)).^2);
        with x(1) = var1 and x(2) = var2, L and E constant values.

Insofar as i ask matlab to optimize a function but some of its variables have to respect a condition with the use of excel database and the use of find function, i have the error message on above.

So I think because of the excel table, the constraint cannot make the fitness function scalar but I don't know how to fix it.

I hope you 've already met this kind of problem and hope you could help me for that.

Thank you!

[Walter Roberson]

We need to see your ObjectiveFunction