nonlinear constrain error must return real value

Hello! When trying to solve an optimization problem( linear and nonlinear constrains) using the ga function it returns the following error:

>> [x,fval] = ga(ObjectiveFunction,nvars,[],[],Aeq,Beq,lb,ub,ConstraintFunction) 
Error using constrValidate (line 59) 
Constraint function must return real value. 
Error in gacommon (line 125) 
[LinearConstr, Iterate,nineqcstr,neqcstr,ncstr] = constrValidate(NonconFcn, ... 
Error in ga (line 363) 
NonconFcn,options,Iterate,type] = gacommon(nvars,fun,Aineq,bineq,Aeq,beq,lb,ub, ... 
    

My non linear constrains are:

function [C,Ceq] = simple_constraint(x)

Ceq=[x(1)*70+x(8)*x(12)-x(3)*x(9); 
     x(2)*70+x(7)*x(11)-x(4)*x(10);
     120-x(3)*(x(11)-x(9));
     120-x(16)*0.2*x(13);
     17.5-x(4)*(x(12)-x(10));
     17.5-x(17)*0.5*x(14);
     90-x(18)*0.5*x(15);
     x(13)-((90-x(9))-(150-x(11)))/(log(abs((90-x(9))/(150-x(11)))));
     x(14)-((264-x(10))-(264-x(12)))/(log(abs((264-x(10))/(264-x(12)))));
     x(15)-((264-70)-(264-100))/(log(abs((264-70)/(264-100))))];
C=[];
end

And my function for minimizing is :

function y = simple_fitness(x)

y=5750*(x(16)^0.59+x(17)^0.59+x(18)^0.59);
end

We tried using the dbstop if error and in the c vector there was a NaN value (we think that is a problem in the 9th nonlean equality constrain, or do the c and ceq equality vectors don't connect?).

Thanks beforehand!

[madhan ravi]

upload the full code where was simple_constraint used?

[Maria Lepouri]

It was used here:

ObjectiveFunction = @simple_fitness;
nvars = 2;    % Number of variables
LB = [0 0];   % Lower bound
UB = [1 13];  % Upper bound
ConstraintFunction = @simple_constraint;
[x,fval] = ga(ObjectiveFunction,nvars,[],[],[],[],LB,UB, ...
    ConstraintFunction)

Answer 1

[Star Strider]

The expression beginning with ‘x(13)’ contains a log argument that could become negative if ‘x(11)’ is greater than 150, and similarly, ‘x(14)’ contains a log argument that could become negative if ‘x(12)’ is greater than 264. Logarithms of negative numbers are complex. Constraining these parameters in the ‘ub’ vector could be one solution, although since I do not know in what order the constraints are checked, might not prevent the error.

To illustrate:

log(abs((90-x(9))/(150-x(11))))
log(abs((264-x(10))/(264-x(12))))

[Maria Lepouri]

I understand what you say but I've already used the abs function to prevent a negative log. I will check it thought and add an abs to both numerator and denominator.