maximizing a function with a nonlinear constraint using fmincon

Dear all,

I have a function 'fun' which I want to maximize with respect to four unknowns: x(1) x(2) x(3) and x(4)

under the constraint

 exp(x(2))+exp(x(3))<0.99

So, I set up something like that

 function [c,ceq] = mycon(x)
 c= exp(x(2))+exp(x(3))-0.99;
 ceq=[];
 end

 x0=[a ;b c; d]  % the initial value
      options=optimset( 'display','off'); % mainly use this
[xx,fval,exitflag,output,lambda,grad,HH]=fmincon('fun',x0,...
 [],[],[],[],[],[],@mycon,  options);

But I get this warning

 Error using mycon
 Too many input arguments.

 Error in fmincon (line 622)
        [ctmp,ceqtmp] = feval(confcn{3},X,varargin{:});

 Error in dokimi_3 (line 96)
 [xx,fval,exitflag,output,lambda,grad,HH]=fmincon('fun',x0,...

 Caused by:
    Failure in initial nonlinear constraint function evaluation. FMINCON cannot continue.

Any ideas what is wrong?

Thank you

[ektor]

Any help? Thank you

[Walter Roberson]

Is it possible that in your actual code that you passed something to fmincon after the options argument?

[ektor]

Thanks Walter

Yes, actually my complete set up is the following

 [xx,fval,exitflag,output,lambda,grad,HH]=fmincon('fun',x0,...
 [],[],[],[],[],[],@mycon ,  options,y,mu, psi,sig20, Vgam,gam0, lam0,Vlam );

where ,y,mu, psi,sig20, Vgam,gam0, are known values that are inserted in the "fun" file

And I get the above error.

Answer 1

[Matt J]

Rewrite mycon to have this form,

   function [c,ceq] = mycon(x, y,mu, psi,sig20, Vgam,gam0, lam0,Vlam )

     c= exp(x(2))+exp(x(3))-0.99;
     ceq=[];

   end

Even if the extra known variables aren't actually used in the computations of the constraints, they have to be there if you use additional arguments to fmincon to pass them in.

Incidentally, passing extra known parameters this way is antiquated. You should use anonymous or nested functions as described here. If you use these alternatives, you will not have to have to pass variables to functions that don't need them.