[Cgl] CglTwomir issue (Re: Success!)

[Robin Lougee-Heimer]

Jean-Paul:

Thanks for the update and being willing to post it in the tracking system. 
 Once it gets in,  I'll assign it to Okay as the PM of the CglTwomir. 

Robin

----------------------------------------------------------------------------------
Robin Lougee-Heimer, PhD
Program Manager, COIN-OR
IBM TJ Watson Research Center
1101 Kitchawan Road, Yorktown Heights, NY 10598
ph: 914-945-3032   fax: 914-945-3434 
robinlh at us.ibm.com
http://www.coin-or.org





"Watson, Jean-paul" <jwatson at sandia.gov> 
12/10/2009 02:38 PM

To
Robin Lougee-Heimer/Watson/IBM at IBMUS
cc
"cgl at list.coin-or.org" <cgl at list.coin-or.org>, Oktay 
Gunluk/Watson/IBM at IBMUS
Subject
RE: CglTwomir issue (Re: [Cgl] Success!)






Hi Robin,
 
It is still a bug; I?ll post it now.
 
Jean-Paul
 
From: Robin Lougee-Heimer [mailto:robinlh at us.ibm.com] 
Sent: Thursday, December 10, 2009 12:31 PM
To: Watson, Jean-paul
Cc: cgl at list.coin-or.org; gunluk at us.ibm.com
Subject: CglTwomir issue (Re: [Cgl] Success!)
 

Jean-Paul: 

If this is still an issue, can you post it in the Cgl bug tracking 
facility?  https://projects.coin-or.org/Cgl/report.  The Project Manager 
for CglTwomir is Oktay Gunluk.   

thanks, 
Robin 

I have isolated a situation in which I believe CglTwomir is returning 
invalid cuts. I have attached a test driver (driver.cpp), a test input 
file (mas74.mps, from miplib2003), and the output trace I obtain on our 
rather vanilla RedHat linux server.


The "catch" is that I'm using CPLEX 11.2, and the OsiCpxSolverInterface. 
Although this technically shouldn't matter, I suppose (the behavior also 
occurs with other versions of CPLEX).

A quick summary of the unexpected behavior:
1)      The LP relaxation for mas74 solves fine - relaxation is around 
10K. Validated with various other solvers.
2)      3 two-mir cuts are found and successfully applied.
3)      Upon re-solve after cut addition, the LP relaxation is something 
like 21K, which is far above the integer optimal of ~11K. Obviously 
incorrect behavior.

Any help is greatly appreciated - I am seeing similar behavior on a 
handful of other test instances.





----------------------------------------------------------------------------------
Robin Lougee-Heimer, PhD
Program Manager, COIN-OR
IBM TJ Watson Research Center
1101 Kitchawan Road, Yorktown Heights, NY 10598
ph: 914-945-3032   fax: 914-945-3434 
robinlh at us.ibm.com
http://www.coin-or.org




"Watson, Jean-paul" <jwatson at sandia.gov> 
Sent by: cgl-bounces at list.coin-or.org 
10/06/2009 11:21 PM 


To
"Matthew Saltzman" <mjs at clemson.edu> 
cc
"cgl at list.coin-or.org" <cgl at list.coin-or.org> 
Subject
Re: [Cgl] Success!
 








That was probably it - the total was around 80K bytes (mas74 is largely 
than I thought). I've attached g'zipped versions of the attachments.

Jean-Paul

--


Dr. Jean-Paul Watson
Discrete Math and Complex Systems Department
Sandia National Laboratories
P.O. Box 5800, MS 1318
Albuquerque, NM 87185-1318
(505) 845-8887
jwatson at sandia.gov




-----Original Message-----
From: Matthew Saltzman [mailto:mjs at clemson.edu] 
Sent: Tuesday, October 06, 2009 8:48 PM
To: Watson, Jean-paul
Cc: cgl at list.coin-or.org
Subject: Re: [Cgl] Success!

How big are your attachments?  They should work, but there's a 40K limit
on message body size.

On Tue, 2009-10-06 at 20:33 -0600, Watson, Jean-paul wrote: 
> Sorry about the posting issues ? I?m not sure why the attachments
> didn?t come through. If someone is able to look at this bug, please
> contact me (I?m not sure who is the CGL Czar) and I?ll forward them
> along.
> 
> Jean-Paul


-- 
               Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs

_______________________________________________
Cgl mailing list
Cgl at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/cgl
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/cgl/attachments/20091210/adcd3fa2/attachment-0001.html