[Ipopt] Ipopt Digest, Vol 96, Issue 4

[Edson Cordeiro do Valle]

  Sean
In your first iteration the objective value is smaller than in the last 
one, but if you check the second column (first iteration), inf_ptr which 
is related to constraints violation, you'll see that your system is not 
satisfying them.
I mean, the solver tries to minimize/maximize the objective but always 
to satisfy the constraints and your constraint are not feasible in the 
first iteration. The behavior you are observing does not necessarily 
means that it converged to a local minimum and seems pretty acceptable.
Regards

                                                  Edson C. do Valle
                                                edsoncv at enq.ufrgs.br
                                                   Skype: edson.cv


>
> Hey everybody,
>
> I have a problem that sometimes ends up increasing the objective and 
> then landing in a local minimum like so:
>
> iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du 
> alpha_pr  ls
>     0  1.2178441e-04 1.05e+01 3.78e-02   0.0 0.00e+00    -  0.00e+00 
> 0.00e+00   0
>     1  8.5210593e+00 6.82e-13 2.54e+04 -11.0 2.21e-01    -  1.00e+00 
> 1.00e+00h  1
>     2  3.7647163e-01 6.82e-13 3.79e+01 -11.0 1.40e-01    -  1.00e+00 
> 1.00e+00f  1
>     3  1.7764708e-01 6.82e-13 7.19e+00 -11.0 2.21e-02    -  1.00e+00 
> 1.00e+00f  1
>     4  1.6117151e-01 6.82e-13 1.36e-01 -11.0 2.75e-02    -  1.00e+00 
> 1.00e+00f  1
>     5  1.6112630e-01 6.82e-13 1.55e-02 -11.0 3.01e-04    -  1.00e+00 
> 1.00e+00f  1
>     6  1.6113003e-01 6.82e-13 3.77e-05 -11.0 3.91e-05    -  1.00e+00 
> 1.00e+00s 47
>
> This eventually kills my solver.
>
> Anybody have any settings advice or tips?
>
> Some of my current settings:
>
> app->Options ()->SetStringValue  ("linear_solver", "ma27");
> app->Options ()->SetStringValue  ("nlp_scaling_method", "gradient-based");
> app->Options ()->SetStringValue  ("linear_system_scaling", "mc19");
> app->Options ()->SetStringValue ("mu_strategy", "adaptive");
>
> Thanks,
> Sean
>
>
> ------------------------------
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>
>
> End of Ipopt Digest, Vol 96, Issue 4
> ************************************
>