"GlobalSearch", for some inputs ends up in "NaN" and sqpInterface and fmincon error

Hi everyone,

I have a question regarding "GlobalSearch". For some inputs, my code sometimes ends up in "NaN" and the code gives this error:

"Error using sqpInterface

Objective function is undefined at initial point. Fmincon cannot continue.

Error in fmincon (line 871)

[X,FVAL,EXITFLAG,OUTPUT,LAMBDA,GRAD,HESSIAN] = sqpInterface(funfcn,X,full(A),full(B),full(Aeq),full(Beq), ...

Error in globaloptim.globalsearch.globalsearchnlp

Error in GlobalSearch/run (line 340)

globaloptim.globalsearch.globalsearchnlp(FUN,X0,A,B,Aeq,Beq,LB,UB,NONLCON,options,localOptions);

Error in Optimization (line 37)

[A,f] = run(gs,problem);

Caused by:

Failure in initial call to fmincon with user-supplied problem structure."

Of course, we all know that the algorithm checks the different inputs and finally goes to the smallest objective function value (error).

Is there a way to ignore the inputs that give "NaN" results and make the algorithm use other inputs to finally reach to lowest objective function value (error)?

[Walter Roberson]

This hints that possibly you might need to tighten your constraints, especially lower bound and upper bound, so that the generated initial values do not fail.

Answer 1

[Umang Pandey]

Hi,

This question has been addressed in the following MATLAB answer:

https://www.mathworks.com/matlabcentral/answers/385848-error-using-barrier-objective-function-is-undefined-at-initial-point-fmincon-cannot-continue

Best,

Umang