[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: [Help-glpk] glpsol.exe differs from AMPL

**From**: |
glpk xypron |

**Subject**: |
Re: [Help-glpk] glpsol.exe differs from AMPL |

**Date**: |
Thu, 28 Jun 2012 05:34:44 +0200 |

Hello Michael,
AMPL is not a solver but an interface to a solver. Hence it is unclear which
solver you are comparing GLPK to, might be IBM Ilog Cplex.
Both GLPK as well as IBM Ilog Cplex will provide **any** of the optimal MIP or
LP solutions if successful. Hence the result is correct.
A model which has a lot of symmetries is very inefficient to solve.
You should try to add additional constraints that force the solution of your
problem to be unique.
Best regards
Xypron
-------- Original-Nachricht --------
>* Datum: Wed, 27 Jun 2012 17:10:47 -0400*
>* Betreff: glpsol.exe differs from AMPL*
>* Hello:*
>* *
>* I am running glpk MIP and in AMPL and finding that the order of the*
>* solutions differ. For example, say I have a model where I need to maximize an*
>* objective based upon paired values X, Y, and Z (and each pairing has a*
>* score). I am not using any additional options for the solutions, eg *
>* specifying*
>* branch&cut, etc.*
>* *
>* AMPL will tell me that the solution is X, Y, Z*
>* glpk will tell me the solution is Z, X, Y.*
>* *
>* I note that the sum of the scores for each solution is equivalent (as it*
>* should be as addition is commutative).*
>* *
>* The problem I am trying to reconcile is is there a physical difference*
>* between ordering objects X, then Y, then Z with the order of Z, then X, then*
>* Y?*
>* *
>* Not sure if this is clear to everyone, but the main point is I see*
>* differences.*
>* *
>* Michael Bramley*
--
NEU: FreePhone 3-fach-Flat mit kostenlosem Smartphone!
Jetzt informieren: http://mobile.1und1.de/?ac=OM.PW.PW003K20328T7073a