[Scip] knapsack-presolve-havoc

[michael.winkler@zib.de]

Hi Matthias,

can you tell me your SCIP version and do you use the default settings? Did
you compile SCIP in debug mode(OPT=dbg)? If not please try. Otherwise can
you maybe add the following lines in cons_knapsack.c after 6524:

SCIPdebug( SCIP_CALL( SCIPprintVar(scip, var, NULL) ) );
SCIPdebug( SCIP_CALL( SCIPprintVar(scip, SCIPvarGetProbvar(var), NULL) ) );

and send me the new output.

It looks like that a fixed to 0 variable is added to the knapsack
constraint?!

Best, Michael

P.S. For bug submission also feel free to use our web page
https://www.edom.mi.uni-erlangen.de/cgi/scipweb/submit.pl .

> Dear SCIPers,
>
> I have a problem with the knapsack constraint. It is caught in an infinite
> loop in presolving. Unfortunately, I cannot reproduce this with a simple
> LP
> file, mainly due to custom constraint handlers.
>
> I attach the output with SCIP_DEBUG in cons_knapsack.c As you can see,
> from
> round 5 on (after the restart, line 3600 in the output log), the same
> variable is added as a lifted item over and over again.
>
> I checked the code with valgrind, but that went fine.
>
> If you need further info, please ask.
>
> Thanks for looking into this.
>
> Best,
> Matthias
>
> --
> Matthias Peinhardt
> Otto-von-Guericke University
> Faculty of Mathematics
> Institute for Mathematical Optimization (IMO)
> Universitätsplatz 2
> D-39106 Magdeburg
> ===================================================
> office: building 02 - room 208
> mail: matthias.peinhardt at ovgu.de
> fon: +49 391 67 11601
> fax: +49 391 67 11171
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>