GlobalSearch and fmincon options

I use GlobalSearch for minizing a function. I use fmincon as the solver. Does using fmincon as the solver mean that I can make use of all options available for fmincon when using GlobalSearch. I wanted to make use of the option 'FunValCheck' using the options specification as

options = optimset('GradObj','off','MaxFunEvals',10000,'Display','on','DerivativeCheck','off','Algorithm','sqp','TolFun',1e-9,'TolX',1e-9,'Display','iter','FunValCheck','on'); 

However, in this case I do not obtain the vector of parameters minimizing the function, which I call theta_hat. I get what is below, whereas if I do not use 'FunValCheck' I obtain the theta_hat.

 Num Pts                 Best       Current    Threshold        Local        Local                 
Analyzed  F-count        f(x)       Penalty      Penalty         f(x)     exitflag        Procedure
 Iter  Func-count            Fval   Feasibility   Step Length       Norm of   First-order  
                                                                       step    optimality
    0           6    2.635536e+05     0.000e+00     1.000e+00     0.000e+00     8.771e+05  
       0        6                                                              -10    Initial Point

GlobalSearch encountered failures in the user provided functions.

The local solver ran once and it failed in a user supplied function.

ans =

          263553.581516057


theta_hat =

     []

[Alan Weiss]

I do not understand what you are trying to do by using the FunValCheck option. Please explain why you want to use it, what you hope to accomplish.

Alan Weiss

MATLAB mathematical toolbox documentation

Answer 1

[Matt J]

The FunValCheck option is intended to abort the optimization if your objective function code does not return a valid value. It appears that that has occurred, just as it was designed.