[Coin-discuss] Re: BCP eats up memory like crazy!
miroslav at andrew.cmu.edu
Mon Jan 2 12:23:41 EST 2006
Thank you, Michael. Limiting the density of the cuts fixes
the problem for the instances I solve.
I wonder how cuts are managed in BCP. What happens after a
cut is discarded from all subproblems where it has been
used? Is the cut removed from data structure or only
pointers to it are deleted?
> Date: Thu, 22 Dec 2005 18:21:50 -0500
> From: Miroslav Karamanov <miroslav at andrew.cmu.edu>
> Subject: [Coin-discuss] BCP eats up memory like crazy!
> To: coin-discuss at list.coin-or.org
> Message-ID: <43AB350E.8030307 at andrew.cmu.edu>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> I am running branch and cut in BCP with aggressive cut generation (dense
> cuts are added too). The program runs out of memory rather quick (2.5GB
> RAM). I find no significant memory leaks.
> I notice that the large memory usage is not caused by the active nodes.
> The number of active nodes decreases at the end of solution but the used
> memory stays high and keeps increasing.
> Also, the size of the subproblems is close to the original size of the
> problem. Even though many cuts are generated, most of them are discarded
> when they become inactive.
> I see no reason for this large memory consumption. The major difference
> to my previous usage of BCP is the aggressive cut generation. (Many
> problems that I used to solve before cannot be completed now on the same
> machine.) Do you have a suggestion why this happens?
> Is there anything in cut management or cut storage that can cause this?
> Thank you very much.
> Message: 2
> Date: Fri, 23 Dec 2005 08:43:57 -0600 (CST)
> From: Michael Hennebry <hennebry at web.cs.ndsu.nodak.edu>
> Subject: Re: [Coin-discuss] BCP eats up memory like crazy!
> To: Discussions about open source software for Operations Research
> <coin-discuss at list.coin-or.org>
> Message-ID: <20051223084032.N47775-100000 at web.cs.ndsu.nodak.edu>
> Content-Type: TEXT/PLAIN; charset=US-ASCII
> On Thu, 22 Dec 2005, Miroslav Karamanov wrote:
>>I am running branch and cut in BCP with aggressive cut generation (dense
>>cuts are added too). The program runs out of memory rather quick (2.5GB
>>RAM). I find no significant memory leaks.
> Methinks that dense cuts might be eating up your memory.
> They take up more memory than sparse cuts,
> both for the cut itself and for the matrix factorizaion.
More information about the Coin-discuss