[SCIP] ZIMPL/ variable and constraint handling

[Ambros Gleixner]

Dear Monia Hamdi,

Your problem of imposing > 0 constraints is a general mathematical 
problem that does not seem specific to SCIP or ZIMPL, so I am afraid 
this mailing list is not the right place to look for a solution.

Although I don't know your formulation, it sounds like it may be 
ill-posed in general (with a sequence of improving solutions converging 
to the optimal zero solution) or at least hard to solve with any solver 
that uses numeric tolerances.

Best regards,
Ambros


On 11.10.2016 23:33, monia hamdi wrote:
> Dear all,
> I wrote a model using ZIMPL. I have 32 variables (17 continuous Y[I] and
> 15 binary X[I])
>
> When I generate the .mps file, I have a message saying I have 32
> variables but when reading the file using scip I get only 17 variables.
>
> This is produced because I have only products when using the variables
> ( Y[i in I]  * X[i in I]) which can be à priori simplified to keep only
> Y[i in I].
> But I need to consider both variables because in practice Y[i in I]  >
> 0.  If my information is correct , ZIMPL imposes the use of  greater or
> equal in the constraints (no strictly greater). In the same time I
> cannot use Y[i in I]  >= epsilon with epsilon near zero.
> I get the all zeros as a solution which is a ''mathematical'' optimal
> solution but not an admissible solution. I don't know how to avoid this
> situation.
>
>
> Thank you for your help.
>
> Monia Hamdi
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>

-- 
Ambros Gleixner, Zuse Institute Berlin, http://www.zib.de/gleixner