[Cbc] Segfaults / wrong infeas

[Gleb Belov]

Hi again,

with revision 2228, there are less failures, and only 1 of them is a
segfault, all others had a bad_alloc or were killed due to no response
after 25 mins (although time limit was 5 mins). The one instance with
segfault is FAILED__noPrepro/CostasArray but I could not reproduce it when
reading from mps.

Among other instances, there are some contradictions in reported results:

----------------------------------------
cntr STATUS: ('../challenge_2015/tdtsp', 'tdtsp.mzn', 'inst_10_34_00.dzn'):
  OPTIMAL:  []
  FEAS:  ['LOG__20151106_130715__CPLEX.res', 'LOG__20151106_130555
__CBC_noPre.res']
  INFEAS:  ['LOG__20151106_130618__CBC.res']
cntr OPTIMAL VALUES: ('../challenge_from2012/mznc12-problems/mspsp',
'mspsp.mzn', 'medium_05.dzn'): {18.0: 'LOG__20151106_130555__CBC_noPre.res',
23.0: 'LOG__20151106_130618__CBC.res'}
cntr OPTIMAL VALUES:
('../challenge_from2012/mznc12-problems/ship-schedule',
'ship-schedule.cp.mzn', '5Ships.dzn'): {482480.0:
'LOG__20151106_130618__CBC.res',
483650.0: 'LOG__20151106_130555__CBC_noPre.res'}
cntr OPTIMAL VALUES: ('../challenge_from2012/mznc13-probs/league',
'league.mzn', 'model15-4-3.dzn'): {290.0:
'LOG__20151106_130555__CBC_noPre.res',
390.0: 'LOG__20151106_130618__CBC.res'}
cntr BOUNDS: ('../challenge_from2012/mznc2014_problems/ship-schedule',
'ship-schedule.cp.mzn', '6Ships.dzn'):
  PRIM: [(607220.0, 'LOG__20151106_130715__CPLEX.res'), (607190.0,
'LOG__20151106_130618__CBC.res'), (607190.0,
'LOG__20151106_130555__CBC_noPre.res')],
  DUAL: [(607265.0, 'LOG__20151106_130715__CPLEX.res'), (607190.0,
'LOG__20151106_130618__CBC.res'), (608250.0,
'LOG__20151106_130555__CBC_noPre.res')]

Again, I invoke Cbc from my code by callCbc("[-preprocess off] -solve
-quit", model), creating the model in memory using OsiClpSolverInterface's
addCols() and addRow(). The mps models of these and the failed instances
are in the updated folder
https://www.dropbox.com/sh/r0k9qony9z32a3k/AACvnaSVFUFBF1NScRkdkIOga?dl=0

Thanks
Gleb

On 28 October 2015 at 03:00, <cbc-request at coin-or.org> wrote:

> Send Cbc mailing list submissions to
>         cbc at list.coin-or.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         http://list.coin-or.org/mailman/listinfo/cbc
> or, via email, send a message with subject or body 'help' to
>         cbc-request at list.coin-or.org
>
> You can reach the person managing the list at
>         cbc-owner at list.coin-or.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Cbc digest..."
>
>
> Today's Topics:
>
>    1. Re: Segfaults / wrong infeas (John Forrest)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 26 Oct 2015 20:09:07 +0000
> From: John Forrest <john.forrest at fastercoin.com>
> To: cbc at list.coin-or.org
> Subject: Re: [Cbc] Segfaults / wrong infeas
> Message-ID: <562E8863.3020406 at fastercoin.com>
> Content-Type: text/plain; charset=windows-1252; format=flowed
>
> Gleb,
>
> In the one I looked at the problem was an innocent (seeming) change.  It
> was in Clp.  If you update using svn or just use the ClpSimplexDual.cpp
> the problem should go away.  If it doesn't then tell me.
>
> John Forrest
>
>
> On 26/10/15 04:56, Gleb Belov wrote:
> > Hi,
> >
> > I have a bunch of models where CBC 2.9.7 produces a segfault or gets
> > ulimit-killed after 25 minutes without any output. For 3 instances,
> > infeasibility is wrongly reported.
> >
> > The models and solver outputs are here:
> >
> https://www.dropbox.com/sh/r0k9qony9z32a3k/AACvnaSVFUFBF1NScRkdkIOga?dl=0
> >
> > Folder FAILED_noPrepro has the models where CBC fails when
> > "-preprocess off" is given to callCbc.
> >
> > Would be good to have those cases corrected as we want to recommend
> > CBC as the default MIP backend in our toolchain
> >
> > Gleb
> > _______________________________________________
> > Cbc mailing list
> > Cbc at list.coin-or.org
> > http://list.coin-or.org/mailman/listinfo/cbc
> >
> >
>
>
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> Cbc mailing list
> Cbc at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/cbc
>
>
> ------------------------------
>
> End of Cbc Digest, Vol 100, Issue 12
> ************************************
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20151109/df8a7e44/attachment.html>