[Clp] Tolerance on column solution value

John J Forrest jjforre at us.ibm.com
Mon Sep 28 13:46:03 EDT 2009


Kish,

In Clp, tolerances apply to the scaled problem so it is possible that it 
was within 1.0e-7.

You can check the secondary status to see if this is the case.

John Forrest





[Clp] Tolerance on column solution value

Kish Shen 
to:
clp
09/28/2009 12:37 PM


Sent by:
clp-bounces at list.coin-or.org







I have been looking at a problem that one of our users reported to me 
while using CLP/CBC (via OSI). The root of the problem is that a 
(non-integer) column has its upper and lower bound set to the same 
value, which is 206.999993<more digits>.

The value assigned to this column after a solve is 207.0, which is 
outside the given bounds, but I assume this is because it is within some 
tolerance -- it is about 1e-5 from 207.0. I assumed that this tolerance 
is specified by OsiPrimalTolerance (I am using OSI), but the
value for this is 1e-7, significantly smaller than 1e-5.

I looked at the OsiClp code, and OsiPrimalTolerance seem to map to 
ClpPrimalTolerance. On other solvers, this does seem to map to the 
correct tolerance, e.g CPX_PARAM_EPRHS for CPLEX. Am I correct? Is
OsiPrimalTolerance the tolerance that a solution value for a column can 
be outside a bound for that column?

Thanks and cheers,

Kish Shen

-- 
This e-mail may contain confidential and privileged material for the
sole use of the intended recipient. Any review, use, distribution or
disclosure by others is strictly prohibited. If you are not the intended
recipient (or authorized to receive for the recipient), please contact
the sender by reply e-mail and delete all copies of this message.
Cisco Systems Limited (Company Number: 02558939), is registered in
England and Wales with its registered office at 1 Callaghan Square,
Cardiff, South Glamorgan CF10 5BT.

_______________________________________________
Clp mailing list
Clp at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/clp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20090928/430af5d6/attachment.html>


More information about the Clp mailing list