[Scip] lp solver clock type (again)
Gregor Hendel
hendel at zib.de
Thu Mar 5 11:49:14 CET 2015
Dear Marko,
you indeed detected a bug in the handling of LP time limits in SCIP. You
could work around this issue by setting a LARGE time limit such as,
e.g., 9 * 1e19. Every double smaller than 1e20 will do. I will send you
a patch in a personal message to resolve this issue correctly.
Thank you very much for diving into this issue and sorry for the
inconvenience,
Gerald, Matthias, and Gregor
Am 05.03.2015 um 09:19 schrieb Horváth Markó:
> Dear all,
>
> I have the same problem as in
> http://listserv.zib.de/pipermail/scip/2015-February/002267.html (and
> http://listserv.zib.de/pipermail/scip/2015-February/002268.html).
> I'm using scip 3.1.1 with soplex 2.0.1. (on 1 thread).
> I want to solve a MIP with 30 seconds time limit (i.e., limits/time
> was set to 30) and with 5% gap limit (i.e., limits/gap was set to
> 0.05). Because SCIP reaches the timelimit but finds no solution
> (actual dualbound is finite (4920.1) and actual primalbound is
> infinite (1e+020) in a minimization problem) I want to resolve (more
> precisely to continue to solve) the problem with 100% gap limit (i.e.,
> I set limits/time to 1e20 and set limits/gap to 1.0) but SCIP prints
> the following message: WARNING: LP solver reached time limit, but SCIP
> time limit is not exceeded yet; you might consider switching the clock
> type of SCIP.
>
> I have found a temporary solution for the problem (I turned off the
> timing of the lp solver, i.e., set SCIP_LPPAR_TIMING to 0), but I hope
> there is an another solution.
>
> Best regards,
> Marko
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
More information about the Scip
mailing list