[Coin-discuss] Extracting the best solution in Cbc.

[acw at ascent.com]

Dr. Forrest, thanks for your response.

The test application, which is provided with Cbc, enables preprocessing by 
default, so we were indeed using it.  (See Cbc/Test/CbcMain.cpp, in the 
function main.)

I turned it off and the results look much better: the results from 
getColSolution and bestSolution are now very similar and adhere to the 
constraints.  The bestSolution vector has some ignorable tiny deviations 
from 0 in some variable; the getColSolution vector seems to lack these. Do 
you have any further advice or clarification?  Without it, I think we will 
go back to getColSolution for now.



John J Forrest <jjforre at us.ibm.com> 
Sent by: coin-discuss-bounces at list.coin-or.org
01/18/2006 08:28 PM
Please respond to
Discussions about open source software for Operations Research 
<coin-discuss at list.coin-or.org>


To
Discussions about open source software for Operations Research 
<coin-discuss at list.coin-or.org>
cc

Subject
Re: [Coin-discuss] Extracting the best solution in Cbc.







Were you using pre-processing?  If not it definitely should have worked. 
If you are then there is now a cleaner way of doing pre-processing using 
CbcStrategy - see Samples/repeat.cpp to give you a clue.  If you that 
fails it can you send me something to help debug - I did some tests and it 
seemed Ok. 

John Forrest 


acw at ascent.com 
Sent by: coin-discuss-bounces at list.coin-or.org 
01/17/2006 05:07 PM 

Please respond to
Discussions about open source software for Operations Research 
<coin-discuss at list.coin-or.org>


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

Subject
[Coin-discuss] Extracting the best solution in Cbc.









(Apologies if this message is a duplication; this message bounced the 
first time because I had not properly joined the mailing list.) 
__________________ 

We are using a COIN/Cbc application that we created by making slight 
modifications to the test example provided with Cbc. 

What is the recommended way to extract the best feasible integer solution? 
 The documentation says to use CbcModel::bestSolution.  We tried this, and 
got bad (infeasible) results.  The example application (in its 
implementation of the SOLUTION command) uses CbcModel::getColSolution, 
which is not documented.  The documentation does mention 
OsiSolverInterface::getColSolution, but this is deprecated (with somewhat 
ambiguous language) in favor of CbcModel::bestSolution. 

We decided to mimic the example, and tried the undocumented 
CbcModel::getColSolution.  This worked nicely with instances that ran to 
completion, but when we added a time limit (with 
CbcModel::setMaximumSeconds) the reported solution was again infeasible. 
(CbcModel::branchAndBound did report finding several integer solutions in 
the running time available.) 

Thank you for your assistance.  We are finding Cbc, overall, quite easy to 
work with and adapt to our needs.
_______________________________________________
Coin-discuss mailing list
Coin-discuss at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-discuss
_______________________________________________
Coin-discuss mailing list
Coin-discuss at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-discuss

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/coin-discuss/attachments/20060119/f973189c/attachment.html>