[Coin-ipopt] Problems using Ipopt

[Andreas Waechter]

Hi Karsten,

Many thanks for your email (yes, English is the preferred language for
this mailing list - and you are doing fine! :)

That Ipopt can't solve your problem might be because something went wrong
in the compilation, or because your model has some bug or is not
well-posed.   I suggest you download a few example problems from the web,
e.g. from Bob Vanderbei's web site at

http://www.sor.princeton.edu/~rvdb/ampl/nlmodels/index.html

such as some Hock and Schittkowski problems at

http://www.sor.princeton.edu/~rvdb/ampl/nlmodels/hs/index.html

and see if Ipopt fails on those as well.  If it does, we need to find out
how you compiled Ipopt (e.g. which operating system, which compiler, what
third party components) and what Ipopt options are you using.  Otherwise,
if it works, you will have to re-write your model.

Let me know how it goes.

Cheers

Andreas

On Thu, 15 Apr 2004, Karsten Theissen wrote:

> Hallo Ipopt-Team especially Andreas,
>
> as we discussed in Dresden I tried to install Ipopt (first in a 32 Bit
> version) and it seemed to be succesfull, because AMPL accepted
> ipoptAMPL as a solver an starts to solve the problem. But even for very easy
> problems i get a
>
> "Regularization parameter getting too large (a): 1.0E24"
>
> error. Maybe there was an error in the compilation ??
>
> Ciao Karsten
>
> P.S.: Is it too impolite to write the problems in german?? It would make it
> much easier???
> P.P.S.: Here I send the complete output from AMPL:
>
> ipopt 2.2.0:
> ******************************************************************************
> This program contains IPOPT, a program for large-scale nonlinear optimization.
>    IPOPT is released as open source under the Common Public License (CPL).
>                For more information visit www.coin-or.org/Ipopt
> ******************************************************************************
>
> Number of variables           :      400
>    of which are fixed         :        0
> Number of constraints         :      299
> Number of lower bounds        :       99
> Number of upper bounds        :        0
> Number of nonzeros in Jacobian:     1092
> Number of nonzeros in Hessian :      200
> get_scale: No scaling of constraints necessary
>
> ITER     ERR        MU       ||C||     ||D||     ALFA(X)  #LS        F
> #cor
>     0 0.998D+00d 0.100D+00 0.990D+00 0.000D+00 0.000D+00   0  0.30197000D+00
> 0
>     0 0.998D+00d 0.200D-01 0.990D+00 0.000D+00 0.000D+00   0  0.30197000D+00
> 0
>  Regularization parameter getting too large (a): 1.0E24
>     1 0.998D+00d 0.990D+00 0.990D+00 0.000D+00 0.000D+00-  0
> 0.30197000D+00193
>  Regularization parameter getting too large (a): 1.0E24
>  solve_barrier: get_step_full returns IERR =  10
>  mainloop: Error: solve_barrier ends with IERR =  10
>
> Number of iterations taken .............                      1
> Final value of objective function is.... 0.3019700000000001D+00
>
> Errors at final point                      (scaled)       (unscaled)
> Final maximal constraint violation is... 0.990000D+00    0.990000D+00
> Final value for dual infeasibility is... 0.997998D+00    0.997998D+00
> Final value of complementarity error is. 0.100000D-01    0.100000D-01
>
> The objective function was evaluated      1 times.
> The constraints were evaluated            1 times.
>
> EXIT: Linear system becomes too ill-conditioned
>
> CPU seconds spent in IPOPT and function evaluations =         40.3300
>
> ipopt 2.2.0: Exit code 10. Check IPOPT output for details.
>
> real     1:51.4
> user       40.3
> sys      1:10.5
>
>
> _______________________________________________
> Coin-ipopt mailing list
> Coin-ipopt at www-124.ibm.com
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-ipopt
>