[SCIP] SCIP multithreading solver DFiberSCIP to solve IP

[Ambros Gleixner]

Hi David,

I feel I have to repeat myself:  Please do not spam this mailing list by 
posting two parts of one question just 15 minutes apart.

Regarding your question: I suppose the SCIP parameter file should have 
the syntax of

    https://scip.zib.de/doc/html/PARAMETERS.php

There you can also find several gap parameters with explanation.

Best,
Ambros




Am 30.04.20 um 06:37 schrieb usa usa:
> Also, I am checking
> https://scip.zib.de/doc-6.0.2/html/PARAMETERS.php
> 
> I would like to set up a mingap (e.g. 2%) so that the solver stops and 
> return a (sub)optimal solution for the given gap.
> 
> thanks
> 
> 
> On Wed, Apr 29, 2020 at 2:28 AM Marc Pfetsch 
> <pfetsch at mathematik.tu-darmstadt.de 
> <mailto:pfetsch at mathematik.tu-darmstadt.de>> wrote:
> 
> 
> 
>     Hi David,
> 
>     let me add to Ambros' answer:
> 
>     If you run a problem in the command line and the problem turns out to be
>     infeasible, you can do "change minuc". This will set up a problem that
>     tries to minimize the number of unsatisfied constraints (i.e., it
>     essentially does what Ambros explained below, but uses indicator
>     constraints).
> 
>      From the output you can sometimes guess what the problem is (if the
>     problem can be solved).
> 
>     Best
> 
>     Marc
> 
>     On 29/04/2020 10:21, Ambros Gleixner wrote:
>      > Hi David,
>      >
>      > I am happy you are so excited about the SCIP Optimization Suite.
>     Because
>      > there are many people subscribed to this mailing list, can you
>     please in
>      > the future make sure to collect all relevant information first and
>      > report them in one e-mail at a time.  Otherwise, we are not even
>     sure,
>      > which question is still relevant.
>      >
>      > Now, e.g., has the infeasibility problem already been resolved?  SCIP
>      > currently has no powerful method to analyze infeasibilities. 
>     However,
>      > you could build your own slack model by either adding
>      >
>      > - a continuous slack variable to each of your suspicious
>     constraints, or
>      >
>      > - convert them into a big-M constraint with an auxiliary binary
>      > activation variable,
>      >
>      > and then minimize the sum of slacks or binaries.  The solution to
>     such
>      > an auxiliary MIP may give you good hints on modeling errors.
>      >
>      > FiberSCIP can currently not be used through the Java interface,
>     but you
>      > could use the SCIPwriteOrigProblem method to write a MPS file of your
>      > Java model, and then solve it on the command line with FiberSCIP, see
>      > the ug/README file in the SCIP Optimization Suite release.
>      >
>      > Best,
>      > Ambros
>      >
>      >
>      >
>      > Am 29.04.20 um 08:36 schrieb usa usa:
>      >> Hi,
>      >>
>      >> I am trying to solve an Integer programming model by running
>      >> java scip from IntelliJ Idea (community 2019.2) on macbook pro.
>      >>
>      >> My java is 11.
>      >> My scipoptsuite: 6.0.2.
>      >> My jscip cheked out from https://github.com/SCIP-Interfaces/JSCIPOpt
>      >>
>      >> My integer programming model took long time to be solved by SCIP
>     solver.
>      >>
>      >> Based on this
>      >> https://scip.zib.de/workshop2014/parascip_libraries.pdf
>      >>
>      >> I would like to find how to use FiberSCIP to improve the
>     performance.
>      >>
>      >> Could anyone help me with this or point out some docs that may be
>      >> helpful?
>      >>
>      >> thanks
>      >>
>      >> David
>      >>
>      >>
>      >>
>      >> _______________________________________________
>      >> Scip mailing list
>      >> Scip at zib.de <mailto:Scip at zib.de>
>      >> https://listserv.zib.de/mailman/listinfo/scip
>      >>
>      >
>     _______________________________________________
>     Scip mailing list
>     Scip at zib.de <mailto:Scip at zib.de>
>     https://listserv.zib.de/mailman/listinfo/scip
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
> 

-- 
Ambros Gleixner, Research Group Mathematical Optimization Methods at 
Zuse Institute Berlin, http://www.zib.de/gleixner