Problem with fmincon: nonlcon

fmincon runs fine without the non-linear constraint given by the function:

function [c,ceq] = DiagCAW11_constraints(theta,RC,r,bsum)

A=diag(theta(11:14),0);

B=diag(theta(15:18),0);

ceq = [];

c = diag(A.^2 + B.^2 - 1);

end

However when running with the constraint, I get the error

"Error using DiagCAW11_constraints
Too many input arguments."

Here is the entire minimization problem:

fmincon('DiagCAW11_likelihood',theta0,[],[],[],[],[],lb,ub,'DiagCAW11_constraints',options,RC,r,bsum);

[Thorbjørn Kristiansen]

As I understand it, the constraint is supposed to have the same amount of input arguments as the function being minimized. Is this a misconception?

[Thorbjørn Kristiansen]

If I add another input variable (x) to the constraint, fmincon runs, but gives the same results as without the constraint.

Answer 1

[Are Mjaavatten]

To transfer the extra inputs to DiagCAW11_constraints you should use an anonymous function:

Make sure that RC, r, bsum are defined. Then define an anonymous function nonlcon as:

 nonlcon = @(theta) DiagCAW11_constraints(theta,RC,r,bsum);

You may transfer parameters to DiagCAW11_likelihood in the same manner:

 likelihood = @(theta) DiagCAW11_likelihood(theta,par1,par2)

Now you may call fmincon:

 theta = fmincon(likelihood,theta0,[],[],[],[],lb,ub,nonlcon,options);

Good luck!