[Coin-ipopt] Lagrange Multipliers in AMPL

Arvind Uppili Raghunathan aur at andrew.cmu.edu
Wed Apr 28 10:31:00 EDT 2004


Dear Karsten,

  In the current AMPL interface for IPOPT we chose not to return the
multipliers but that can be fixed easily. Andreas can easily fix that in
the next release.

Cheers
Arvind

On Wed, 28 Apr 2004, Karsten Theissen wrote:

> Hello IPOPT-Experts,
>
> after some problems in the beginning, everything is working now! Everything??
> Of course not ;-), but i think it is only a small problem.
>
> I test all my old and new AMPL-LOQO problems with IPOPT (and by the way IPOPT
> is doing a great job...) but there are some problems with the Lagrange
> multipliers. When I was working with LOQO you can get them by evaluating the
> name of your constraints e.g. with
>
> s.t. dgl1{i in 0..Nt-1}:
>
> you get them with
>
> dgl1[i]     i=1..Nt-1
>
> I think with IPOPT it should be the same but there are all equal to zero.
> Maybe it is only a question of a missing ipopt_option.
> It would be nice if anybody could help.
>
> Greetings
>
> Karsten Theißen
>
> _______________________________________________
> Coin-ipopt mailing list
> Coin-ipopt at www-124.ibm.com
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-ipopt
>
>





More information about the Coin-ipopt mailing list