[Scip] Column Generation: Display columns
Dimitri Sverdlov
dimitri.sverdlov at inria.fr
Thu Dec 3 03:01:42 MET 2009
Thanks a lot, Timo, that helps!
Dimitri
Timo Berthold wrote:
> Hi Dimitri.
>
> What you can do, is the following:
> 1) If you are just interested in the actual solution of the coloring problem,
> you can write it by calling
> SCIP> write problem foo.csol
> which shows sou the color classes (by corresponds to the columns).
> 2) YOu couls also stop SCIP by some limit, e.g.
> SCIP> set limits solutions 1
> or
> SCIP> set limits nodes 1
> and write the MIP relaxation afterwards:
> SCIP> write mip foo.lp 0 1
>
> I hope this help a bit.
> Best regards,
> Timo
>
>
> E.g., for the problem myciel3 and a solution limit of one, this gives rise to
> the following two files:
>
> myciel3.csol
> -------------------
> myciel3 4 generated by ColumnGenerationColoring
> 0 1 0 1 2 2 1 3 1 2 0
> -------------------
>
>
> myciel3.lp
> -------------------
> Minimize
> Obj: +1 _var0_ +1 _var1_ +1 _var2_ +1 _var3_
>
> Subject to
> Node-Constraint1: -1 _var0_ -1 _var2_ -1 _var3_<= -1
> Node-Constraint2: -1 _var1_<= -1
> Node-Constraint3: -1 _var0_ -1 _var3_<= -1
> Node-Constraint4: -1 _var1_<= -1
> Node-Constraint5: -1 _var2_<= -1
> Node-Constraint6: -1 _var2_ -1 _var3_<= -1
> Node-Constraint7: -1 _var1_<= -1
> Node-Constraint8: -1 _var3_<= -1
> Node-Constraint9: -1 _var1_<= -1
> Node-Constraint10: -1 _var2_<= -1
> Node-Constraint11: -1 _var0_<= -1
>
> Bounds
> -0<= _var0_
> -0<= _var1_
> -0<= _var2_
> -0<= _var3_
>
> Generals
> _var0_ _var1_ _var2_ _var3_
>
> End
> -------------------
>
> Am Thursday 03 December 2009 01:33:00 schrieb Dimitri Sverdlov:
>
>> Hello,
>>
>> I'm working with the Coloring example from SCIP.
>> I wounder how it is possible to display the solution with the
>> interactive optimizer shell in case of column generation.
>> That is, display solution shows the values of the variables, but of
>> course I also want to see the columns corresponding to the variables. Is
>> that possible?
>>
>> Best regards,
>> Dimitri
>>
>
>
More information about the Scip
mailing list