[Coin-lpsolver] (no subject) (should be: possible slow solving with CLP)
John J Forrest
jjforre at us.ibm.com
Mon Aug 22 10:56:32 EDT 2005
Kish,
It would probably affect both. When I got close to your error it was not
in a loop it was just that a free variable was at 1.0e20 which blows
precision.
John Forrest
Kish Shen <ks15 at icparc.ic.ac.uk>
Sent by: coin-lpsolver-bounces at list.coin-or.org
08/22/2005 10:46 AM
To
coin-lpsolver at list.coin-or.org
cc
Subject
Re: [Coin-lpsolver] (no subject) (should be: possible slow solving with
CLP)
John,
> Thank you for finding the weakness (a free variable entered the basis
not
> quite as I expected)
Do you expect this to affect both the testit run I tried (which seemed to
have got into a loop), and the original run with the Osi interface (which
does not appear to be in a loop)?
In the original run, where the problem is set up using the Osi interface
rather than reading from the MPS file, the code I am running alway
give bounds to variables -- it initially assigns -infinity to infinity (or
what getInfinity() returns) as bounds to the variable, and then adjust the
bounds for specific variable(s) if they are available.
This is probably unusual for Mathematical Programming, where I think the
default bounds for variable is normally 0 to infinity. Could this be a
problem for the solving of l30?
Would you like me to generate the Osi calls for my run of the l30 problem
instance?
Cheers,
Kish Shen
IC-Parc
_______________________________________________
Coin-lpsolver mailing list
Coin-lpsolver at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-lpsolver
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20050822/53c0e5a3/attachment.html>
More information about the Clp
mailing list