Changing tolCon in fmincon, solves for tolCon 1e-1 but not for 1e-2

Hi,

My question relates to finding a feasible solution with fmincon by changing the constraint tolerance.

My optimization problem will solve and find an optimal solution if I set tolCon to 1e-1, however if it is 1e-2 or smaller it will not find a feasible solution.

Would the optimal found be accurate?

Thanks in advance for any advice

Answer 1

[Alan Weiss]

It sounds as if the "optimal solution" fmincon finds with a 1e-1 TolCon is not feasible to a small tolerance. Or maybe it is--you gave us very little information.

So, is the solution really feasible? Did you evaluate the constraints?

Alan Weiss

MATLAB mathematical toolbox documentation