[Coin-discuss] CBC "standard" solver

[Brady Hunsaker]

Thanks, John.

One of the projects I'm working on is to try to predict the time to
solve an instance, building on work by Cornuejols, Karamanov, and Li.
I'm going to have a student look to see if we can do so usefully with CBC.

I'm also working on a method to determine good option settings without
trying every combination (right now using a neural net).  If this works
out, my hope is that we can use it to determine good default settings
for CBC whenever a new feature or cut is added.

I'll keep you posted on both of these.

Brady

John J Forrest wrote:
> 
> Brady,
> 
> I think CBC is now robust enough to go into production so I hope to get
> the library to version 1.0 by INFORMS.  At present it is at 0.99!
> 
> I have also created a stand alone CBC executable which shares code with
> my stand  alone CLP executable and this is what people should use.   It
> is intended to give good performance with no options but there are many
> options.  It is intended to give hints in the same way as CLP although
> the explanations need more work.  Individual cuts may be switched on or
> off and you can also switch on all (well most) cuts at root, in tree or
> let the code decide.
> 
> To make it more useful there is also a savesolution command which saves
> the solution to file in a very simple format which can be used by other
> programs.  The solution command from clp is also there to print solutions.
> 
> To obtain this executable (solve) use "make solve" in the Cbc directory.
>  I will be working on this to make it better so any suggestions are
> welcome.
> 
> John
> 
> 
> *Brady Hunsaker <hunsaker at engr.pitt.edu>*
> Sent by: coin-discuss-bounces at list.coin-or.org
> 
> 10/30/2005 08:07 PM
> Please respond to
> Discussions about open source software for Operations Research        
> 
> 
> 	
> To
> 	coin-discuss at list.coin-or.org
> cc
> 	
> Subject
> 	[Coin-discuss] CBC "standard" solver
> 
> 
> 	
> 
> 
> 
> 
> 
> Is there a general CBC-based solver that can be easily used without coding?
> 
> I am doing some work on predicting time to solve and separate work on
> tuning parameters and would like to work with a general CBC solver.
> 
> I know that CBC is intended primarily as a library, but I believe we
> should also have one or more general solvers clearly indicated for
> people who don't need/want to code their own.  I thought there was one,
> but it wasn't obvious to me as I looked through the files.
> 
> Thanks,
> 
> Brady
> _______________________________________________
> 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


-- 
Brady Hunsaker
Assistant Professor
Industrial Engineering
University of Pittsburgh
http://www.engr.pitt.edu/hunsaker/