[Scip] Regarding a MIP problem

[gaurav khanna]

Hi Tobias,

Thanks for all your help. I guess the results you are getting is by using
the ILOG CPLEX in conjunction with SCIP. Since I do not have a license to
it, Therefore I am tried SCIP both with SOPLEX 1.3 and an academically
licensed version of MOSEK on another problem instance. For, both of these,
the problem instance has been running for 3 hours now even after adding all the changes that you
had suggested to scip.set.

The problem instance is present at
http://www.cse.ohio-state.edu/~khannag/problem_data.8_40_44.mps

Regards
Gaurav

Gaurav Khanna
Phd Student
CSE Department,OSU

Phone (office):614-292-7036

On Fri, 8 Dec 2006, Tobias Achterberg wrote:

> Hi Gaurav,
>
> I think it should be useful to remove the integrality restriction on the integer
> objective variable. It will be integral anyway for every solution that is LP optimal
> and integral for the binary variables.
>
> Unfortunately, SCIP is not able to detect this implied integer property. Thus, it
> might unnecessary branch on this variable or create cuts enforcing the integrality
> and so on. I removed the integrality restriction, and the root node bound gets to 3
> (the optimal value), which was not the case with the integrality restriction.
>
> For this, you should disable strongcg cuts; they only seem to hurt performance:
> separating/strongcg/freq = -1
>
> cmir-cuts, however, seem to be slow but useful to find the 3 in the root node:
> separating/cmir/freq = 0 (the default setting)
>
>
> Tobias
>
>
>
>
> gaurav khanna wrote:
> > Hi all,
> >
> > I have tried to use SCIP with MOSEK to solve a 0-1 MIP problem. The
> > problem has 54061 variables and 78326 constraints. The instance has been
> > running for the last 3 hours using the standard parameter file scip.set.
> > To reduce the running time of the solver, is there any way I can change
> > the parameter values in the scip.set file to speedup the  solution.
> > Are there any specific parameters which I should target? Also,
> > I am ok with getting feasible solutions to the MIP which need not be
> > globally optimal. Is there a way to do that using SCIP that is to get a
> > feasible solution in a lesser amount of time without significant loss in
> > quality as compared to the global optimal.
> >
> > The problem instance (MPS) is present at
> >
> > http://www.cse.ohio-state.edu/~khannag/prob_8_lc.mps
> >
> > Regards
> > Gaurav
> >
> >
> > Gaurav Khanna
> > Phd Student
> > CSE Department,OSU
> >
> > Phone (office):614-292-7036
> >
> > _______________________________________________
> > Scip mailing list
> > Scip at zib.de
> > http://listserv.zib.de/mailman/listinfo/scip
>