[Scip] Optimal duals in *.sol returned by SCIPAMPL

[Ambros Gleixner]

Dear Vladimir,

sorry for the wait.  There is currently no AMPL-SoPlex interface and we 
were not planning to implement this.  So the fastest way would probably 
be to make LP duals available in SCIPAMPL if presolving is turned off.

Stefan and me are currently too busy to do this, but we are happy to 
support you if you want to go ahead.

Ambros




Am 12.08.2014 um 18:46 schrieb Vladimir VV:
> Dear Stefan,
> I tried to use SCIPAMPL as an "AMPL-interface" for SoPlex.
>
> We did the same for CBC (Coin-Branch-and-Cut) MILP solver wich is
> based on CLP (Coin-LP) solver, i.e. for "continues" LP-problem the
> command
> cbc stub.nl -AMPL wantsol=7
> gives stub.sol with optimal duals "inside".
>
> At https://github.com/ampl/coin/tree/master/Osi/src/OsiSpx
> I see Coin_OSI-intrerface for SoPlex.
> May be AMPL-interface for SoPlex exists somewhere ?
>
> Vladimir.
>
> On Tue, Aug 12, 2014 at 6:51 PM, Stefan Vigerske
> <stefan at math.hu-berlin.de> wrote:
>> Hi,
>>
>> Duals in terms of single values associated with variable bounds and
>> constraints make only sense for convex problems, which isn't SCIPs domain.
>> One could get them through the AMPL interface when solving a LP without
>> presolving with SCIP, but I doubt the effort is worth it. Just use an LP
>> solver for such cases.
>>
>> Stefan
>>
>>
>> On 08/12/2014 04:10 PM, Vladimir VV wrote:
>>>
>>> Dear all,
>>> Is it possible to get optimal dual variables by *.sol file returned by
>>> SCIPAMPL ?
>>> I see following operator
>>> write_sol((char*)msg, x, NULL, NULL);
>>> in line #1386 of
>>> .../scip-3.1.0/interfaces/ampl/src/reader_nl.c
>>>
>>> So, nothing duals (the first NULL) nor "wantsol" options ?
>>> What is the reason of that ?
>>> BTW, the same was for scip-3.0.2...
>>>
>>> Thanks in advance.
>>> Vladimir V. Voloshinov.
>>> _______________________________________________
>>> Scip mailing list
>>> Scip at zib.de
>>> http://listserv.zib.de/mailman/listinfo/scip
>>>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>

-- 
____________________________________________________________
Ambros M. Gleixner
Zuse Institute Berlin - Matheon - Berlin Mathematical School
http://www.zib.de/gleixner