[Coin-lpsolver] Problem with COIN MIP solver

[John J Forrest]

Ricky,

37668.9 is a better solution than 39361.67!  It is also the continuous
solution value for this problem, so I am not sure what happened.  When I
run using the stand-alone solver (in many ways) it always gives 39361
answer.

Also I can not give any advice on cut generators as it solves in zero nodes
even without cut generators.  The general answer is to run using the
stand-alone solver and see which cut generators it said were giving cuts.

John Forrest


                                                                           
             "Ricky O'Brien"                                               
             <rickob at internode                                             
             .on.net>                                                   To 
             Sent by:                  <coin-lpsolver at list.coin-or.org>    
             coin-lpsolver-bou                                          cc 
             nces at list.coin-or                                             
             .org                                                  Subject 
                                       [Coin-lpsolver] Problem with COIN   
                                       MIP solver                          
             11/17/06 07:26 AM                                             
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           




We have been running the branch and cut (Cbc) algorithm on a variety of MIP
problems with both a commercial LP/MIP package and COIN.  For most
problems, both solvers produce the same answer, but for the attached file
(a zipped mps file), we believe that the COIN branch and cut algorithm is
giving us a less than optimal solution.  With the commercial package, the
objective value that we get is 39361.67, however COIN is giving an
objective value of 37668.9 .





We are also interested in improving the performance of the solver for our
applications, and, we have noticed the COIN-OR cut generation library
(Cgl).  As this is not my area of expertise, l was wondering if there is a
general rule of thumb to guide me in the choice of which cut generator to
use for MIP problems where there are a moderate number (a few hundred) of
binary variables?





Thank you for any help.


Ricky O’Brien[attachment "2006-01-01 0330.zip" deleted by John J
Forrest/Watson/IBM] _______________________________________________
Coin-lpsolver mailing list
Coin-lpsolver at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-lpsolver