[Coin-discuss] Cbc, Symphony, and GLPK

[Menal Guzelsoy]

Jen-Min,

How do you check the violation of the constraints? Are you using the
solution displayed on the screen? SYMPHONY prints out the solution for
continuous variables with 3 fractional digits, the last of which is
obviously rounded. In this problem, it might not be that accurate to
use to check the violation of constraints.

Menal.

Menal Guzelsoy
Ph.D. Candidate
Industrial and Systems Engineering Dept.
Lehigh University
1610 4620455



On Wed, Jun 3, 2009 at 3:19 AM, John J Forrest <jjforre at us.ibm.com> wrote:
> For Cbc I am unable to reproduce problem on Linux. driver4 and stand-alone
> cbc with various option permutations all gave the answer of -
>
> BINV0001, 04, 10 and 14 at 1 and rest at 0.
>
> which when the problem with all binary variables fixed is solved does not
> seem to violate the constraints. Which constraints do you think that
> solution violates?
>
> John Forrest
>
> "Cheng,Jen-Min" ---06/02/2009 01:32:33 PM---Microsoft Visual Studio 2008 has
> been used to generate executables for Cbc driver4 2.3, Symphony 5.2, and
> GLPK 4.38. These e
>
>
> From:
> "Cheng,Jen-Min" <jcheng at mwdh2o.com>
> To:
> <coin-discuss at list.coin-or.org>
> Cc:
> "Louie,Peter W" <plouie at mwdh2o.com>
> Date:
> 06/02/2009 01:32 PM
> Subject:
> [Coin-discuss] Cbc, Symphony, and GLPK
> ________________________________
>
>
>
> Microsoft Visual Studio 2008 has been used to generate executables for Cbc
> driver4 2.3, Symphony 5.2, and GLPK 4.38. These executables can solve many
> MIPLIB 2003 problems. Cbc driver4 2.3 did outperform other two solvers in
> solving these MIPLIB 2003 problemsr as Cbc developers had claimed.
>
> An artificial example of water supply and allocation had been created and
> tested with Cbc driver4, Symphony, and GLPK. GLPK produced a feasible
> solution. Symphony 5.2 produced a feasible solution, which violated several
> constraints. Cbc driver4 2.3 generated a feasible solution, which violated
> more constraints than Symphony 5.2.
>
> The mps file is attached. Will any please find out why Cbc and Symphony
> cannot produce a feasible solution.
>
> Thanks for your time and help.
>
> [attachment "test100.mps" deleted by John J Forrest/Watson/IBM]
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>
>
>
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>
>