[Cbc] Huge constraint file

[Yves Touchard]

Hi andrew,

Thanks for your message,

With diving options, the problem is solved in 7 minutes (454 sec). These options seems to be very efficient!
I just don't know if i can use them for any kind of problem and if not, how to decide to use them or not.

For warm starts, i have actually no idea how i can constrain our problems. I will think about it.

Thank again for your help.

Yves

On 11/04/2014 03:41 AM, Andrew Wald wrote:
Hi Yves,

I don't claim to be an expert with Cbc, but I used a couple techniques that were successful in my last project.

I had a pretty large workforce optimization problem that was wholly intractable with Cbc; it took forever to reach even a 5% optimality gap with CPLEX. Apart from throwing more computer at the problem, I found the following to be helpful:

1) Try experimenting with different combinations of diving options. I ran a bunch of tests and found that DivingVectorlength, by itself, was the most effective for my formulation.
2) Try warm starts. I overly constrained my problem to something Cbc could solve relatively quickly, which I then plugged back in as a feasible integer starting point for the next run (where those constraints were relaxed). This is what took me from intractability to something that Cbc could churn through.

I hope this helps with your particular problem,
Andrew Wald



On Mon, Nov 3, 2014 at 5:02 AM, Yves Touchard <ytouch at dxo.com<mailto:ytouch at dxo.com>> wrote:
Hello,

I have a (huge, more than one million of lines...) constraint file (https://www.dropbox.com/s/6y7oid24ecv8oh7/huge.lp?dl=0) taking a long time (around 4 hours) to be solved (Log file: https://www.dropbox.com/s/rx7m8xtzbe9wsd8/huge.log.txt?dl=0)
I understand that such a huge file could be tricky for the solver. So, what is the best solution:

  *   Doing a pre-analysis to split the problem in two or more sub-problems.
  *   Use specific options to reduce elapsed time (in a deterministic way - such as gap). I tried with diving options but there was a memory allocation error (https://www.dropbox.com/s/rd00y6mfapedq17/hugeErr.log.txt?dl=0)
  *   Let you search and find if there is a way to speed up the coin process.

Thanks and regards,

Yves Touchard

------------------------------------------------------------------------------
This message and any attachments (the "message") are confidential and intended solely for the addressee(s). Any unauthorized use or dissemination is prohibited. E-mails are susceptible to alteration. Neither DxO Labs nor any of its subsidiaries or affiliates shall be liable for the message if altered, changed or falsified.
Ce message et toutes les pièces jointes (ci-après le "message") sont confidentiels et établis à l'intention exclusive de ses destinataires. Toute utilisation ou diffusion non autorisé est interdite. Tout message électronique est susceptible d'altération. DxO Labs et ses filiales déclinent toute responsabilité au titre de ce message s'il a été altéré, modifié ou falsifié.

_______________________________________________
Cbc mailing list
Cbc at list.coin-or.org<mailto:Cbc at list.coin-or.org>
http://list.coin-or.org/mailman/listinfo/cbc




------------------------------------------------------------------------------
This message and any attachments (the "message") are confidential and intended solely for the addressee(s). Any unauthorized use or dissemination is prohibited. E-mails are susceptible to alteration. Neither DxO Labs nor any of its subsidiaries or affiliates shall be liable for the message if altered, changed or falsified.
Ce message et toutes les pièces jointes (ci-après le "message") sont confidentiels et établis à l'intention exclusive de ses destinataires. Toute utilisation ou diffusion non autorisé est interdite. Tout message électronique est susceptible d'altération. DxO Labs et ses filiales déclinent toute responsabilité au titre de ce message s'il a été altéré, modifié ou falsifié.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20141104/d062339d/attachment.html>