Why feasibility always zero? what does it mean? thx

I use fmincon to do the optimization:

 End diagnostic information
                                            First-order      Norm of
 Iter F-count            f(x)  Feasibility   optimality         step
    0       5    4.405383e-01    0.000e+00    7.795e-01
    1      10    6.294627e-02    0.000e+00    1.920e-01    1.258e+00
    2      15    2.902131e+02    0.000e+00    2.095e+01    8.519e+01
    3      20    2.590246e+02    0.000e+00    1.978e+01    5.210e+01
    4      25    4.631491e+02    0.000e+00    2.644e+01    2.449e+02
    5      30    4.482603e+01    0.000e+00    8.201e+00    3.028e+02

But feasibility always zeros, what does it mean? many thanks

Answer 1

[Matt J]

But feasibility always zeros,

They are zero only to the 3 decimal places displayed on the screen.

[Zhixiao]

so, my situation is normal?

[Matt J]

I mean that there's not anything obviously wrong with what you've shown. If fmincon detected a problem it would tell you and indicate so in the exitflag output.

[John D'Errico]

+1. I had to laugh, as this comment made me recall one of my favorite poems, the Unknown Citizen, by W.H. Auden.

"Was he free, was he happy? The question is absurd:
Had anything been wrong, we should certainly have heard."

In a sense, this is true of fmincon. We have no reason to expect that anything is wrong from what we see. In fact, the output shown seems perfectly reasonable. But fmincon does not always know that it has gotten into trouble. You need to look at the result. Does it make sense to you?

Answer 2

[Alan Weiss]

You can find the meaning of the fmincon iterative display here. You will see there that feasibility 0 means that the current point is feasible, meaning that all constraints are satisfied at that point.

Alan Weiss

MATLAB mathematical toolbox documentation

[Matt J]

Hi Alan,

I think it would be helpful to have links to that page in the section of the fmincon documentation that describes the Display option. Same thing for other solvers.

[Alan Weiss]

That's a good idea, Matt. I'll get on it.

Alan Weiss

MATLAB mathematical toolbox documentation