[Coin-lpsolver] Adding a new condition

John J Forrest jjforre at us.ibm.com
Sun May 8 08:06:57 EDT 2005


David,

What you are doing is cleaner and more predictable.  If you want to you 
can just set integers by model.solver()->setInteger(sequence).  But doing 
branch and cut may have set other things e.g. it is possible that integer 
variables which had naturally been at their bound may or may not have been 
fixed.  One thing that you would certainly have to do is reset the integer 
cutoff to infinity as otherwise the next branch and bound will say it 
can't find a better solution.

I will probably clean up some of these less obvious aspects before release 
1.0.

John Forrest



svoboda at fi.muni.cz 
Sent by: coin-lpsolver-bounces at list.coin-or.org
05/06/2005 05:50 AM

To
coin-lpsolver at list.coin-or.org
cc

Subject
[Coin-lpsolver] Adding a new condition






Hello,

I've been using CbcModel class as a tool for solving the integer 
programming problem. I do the iterative computation in the sence that:

- the first iteration solves traditional linear programming problem 
(without any integer variables)

- before the second iteration starts, some continuous variables are set to 

integer variables.

- keeping this course in each iteration, some variables are set to 
integer until a specific solution is obtained.

The question is: may I add these "integer limitations" to CbcModel 
directly (not to construct this model in each iteration), or is it 
necessary to add this type of condition to CoinModel and then to construct 

a new CbcModel?


With regards

David Svoboda

----------------------
Faculty of Informatics, Masaryk University 
Botanická 68a, 602 00 BRNO, Czech Republic
e-mail: svoboda at fi.muni.cz
tel: +420-549 494 383_______________________________________________
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/20050508/24f7f54c/attachment.html>


More information about the Clp mailing list