Primal Infeasibility when adding objective function

[giovan...@gmail.com]

Dear Mosek group members,

When implementing a new LMI problem, in case only the constraints are implemented, I get the message that the problem was successifully solved. On the other hand, when I ask for an optimal solution (-logdet(X)), the solver gives me a message of unbounded objective function and I get a primal infeasible certificate. Even by constraining in all ways possible I still get the same message. Is there a "numerical' reason for this? Even in case I have an unboundd problem I would expect a dual infeasible certificate....

Thank you all very much for your time.

Best regards,

Gio  

[Michal Adamaszek]

Hi,

How do you add this objective in Mosek? Can you write some details about your code and mosek version?

Michal

[Michal Adamaszek]

Hi again,

Are you using Yalmip? Remember that Yalmip dualizes, so if Mosek says primal infeasible, then your model is dual infeasible (according to the message you get). So that would all look perfectly OK.

Michal

On Monday, December 17, 2018 at 4:49:31 PM UTC+1, giovan...@gmail.com wrote:

[giovani caneda]

Oh, okay then!

Thank you very much. I am still trying to add various constraints but cant seem to find a way to constraint it properly. Is there a way to find which variable is unbounded?

Gio

On Mon, Dec 17, 2018 at 5:04 PM giovani caneda <giovan...@gmail.com> wrote:

Hi Michal,

I am asking for mosek to: min -geomean(X) (almost equivalent to logdet in yalmip)

s.t.F = [F,[(1-l0)*X,zz,(X'*A')+Y'*B';zz',mu*eye(6),eye(6)';A*X+B*Y,eye(6),X]>=(1e-06)*eye(18)];

I am using matlab 2017b and yalmip to write the optimization function.  My mosek version is  8.1.0.45, if I am not wrong.

Thank you for your time,

Kind regards,

Gio

--
You received this message because you are subscribed to the Google Groups "mosek" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mosek+un...@googlegroups.com.
To post to this group, send email to mo...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[giovani caneda]

Hi Michal,

I am asking for mosek to: min -geomean(X) (almost equivalent to logdet in yalmip)

s.t.F = [F,[(1-l0)*X,zz,(X'*A')+Y'*B';zz',mu*eye(6),eye(6)';A*X+B*Y,eye(6),X]>=(1e-06)*eye(18)];

I am using matlab 2017b and yalmip to write the optimization function.  My mosek version is  8.1.0.45, if I am not wrong.

Thank you for your time,

Kind regards,

Gio

On Mon, Dec 17, 2018 at 4:52 PM Michal Adamaszek <michal.a...@mosek.com> wrote:

--

[Michal Adamaszek]

The Yalmip forum will be more suitable for these questions.

Best,

Michal

[giovani caneda]

Ok, thank you very much!

--