I am using fmincon to optimize the result of a unknown executable function (function.exe).
Fmincon changes the variables in the size of 1e-3 ~ 1e-4. But the result of the function leads to the same result. I'm sure that the changes are within the rounding tolerance of output of the function. How can i change the settings of fmincon to use a larger step size.
I already tried 'OptimalityTolerance', 'FiniteDifferenceStepSize' and 'DiffMinChange'. Non of them changed the stepsize of the variation of the variables.