[Cbc] Pre-Processing Bug

[Haroldo Santos]

Hi,

I think I've found a bug in pre-processing phase of CBC.

What I was doing (and worked flawlessly for all MIPs except this one) was
to inform to CBC a feasible solution found using a greedy heuristic. This
solution was informed considering variables of the pre-processed model
(checking for the original variables in originalColumns() ).

What happened is that the pre-processed model for this case included a new
set of constraints (expected) which make this valid solution to be
considered infeasible (unexpected). When I fix the same set of variables in
the original LP there is no problem.

To make it easy to check for this problem I've wrote a small code which
fixes the same variables in the original and in the pre-processed problem
and made it available here:

http://www.decom.ufop.br/haroldo/files/ppBug.tar.gz

In infRows.txt I listed tree constraints that when removed allows
feasibility to be obtained again.

Thanks in advance,

Haroldo

-- 
=============================================================
Haroldo Gambini Santos
Computing Department - Universidade Federal de Ouro Preto - UFOP
email: haroldo [at ] iceb.ufop.br
home/research page: www.decom.ufop.br/haroldo/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20120120/4d295bcc/attachment.html>