[Ipopt] empty x

Victor Fachinotti vfachino at intec.unl.edu.ar
Wed Dec 14 09:02:13 EST 2016


Hello colleagues,

I have noticed something wrong during a call to iterfunc using IPOPT under
Matlab. At a given intermediate iteration (the 56th, neither the first nor
the last one), the x field in the third argument of iterfunc is empty.
However, the solver goes on.

 

Here is the evolution of the optimization:

 

iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du alpha_pr
ls

  50 2.2029586e+001 2.45e-002 1.36e+000  -8.6 8.48e-002    -  8.20e-006
2.58e-001w  1

In iteration 50, 1 Slack too small, adjusting variable bound

  51 9.1022741e+001 1.93e+000 1.44e+000  -8.6 2.75e-002    -  1.00e+000
2.74e-001w  1

  52 9.1022741e+001 1.91e+000 1.46e+000  -8.6 5.41e-001    -  4.48e-001
9.72e-003w  1

  53r3.0398264e+000 1.35e-003 9.99e+002  -2.9 0.00e+000    -  0.00e+000
4.92e-007R 19

  54r3.9243117e+000 2.01e-003 1.31e+002  -9.0 1.35e+000    -  8.69e-001
3.87e-007f  1

  55r1.2696788e+000 1.46e-009 6.82e+001  -9.0 1.75e-003    -  4.80e-001
8.60e-001f  1

  56r1.1584010e+000 6.04e-010 1.72e+001  -9.0 3.51e-003    -  7.47e-001
5.49e-001f  1

  57r1.2401559e+000 2.51e-011 6.86e+000  -9.0 1.94e-003    -  6.02e-001
9.73e-001f  1

  58 1.2401253e+000 1.83e-011 2.61e-001  -8.6 1.23e-003    -  9.01e-001
2.74e-001h  2

  59 1.2397894e+000 1.22e-011 1.78e-001  -8.6 2.76e-002    -  1.00e+000
3.34e-001h  1

 

And here are the inputs of iterfunc:

 

iter =    56

fobj =  1.158400955836191

x = 

                      x: []

                 inf_pr: 0.002513493667610

                 inf_du: 17.243376684535178

                     mu: 9.999999999999999e-10

                 d_norm: 0.003511158700339

    regularization_size: 0

               alpha_du: 0.747343537290606

               alpha_pr: 0.549440501968642

              ls_trials: 1

 

 

So, what is actually the x the objective and the other variables were
computed for?

 

Thanks,

 

Víctor D. Fachinotti

CIMEC (UNL/CONICET)

Tel: 0342 4511594 interno 7045

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20161214/4c819bed/attachment.html>


More information about the Ipopt mailing list