[Clp] [Coin-discuss] Problems with cbc - bad dual

Ted Ralphs ted at lehigh.edu
Sun Feb 14 12:34:55 EST 2010


The message is coming from Clp, the underlying linear programming
solver, not from Cbc itself. It appears Clp is having numerical
difficulties, perhaps due to an ill-conditioned constraint matrix.
I've copied this reply to the clp and cbc mailing lists to continue
the discussion. If you can post some more information, such as what
version of Cbc you are using, what platform you are using it on, and
perhaps an input file on which we can replicate this problem, we'll
try to help.

Cheers,

Ted

On Sun, Feb 14, 2010 at 11:40 AM, Mauro Scarioni
<mauro_scarioni at it.ibm.com> wrote:
>
> Hy guys
>
> I have a problem with a Coin/cbc application.
>
> During the execution of some instance at a certain point the solver seems
> hold (but the cpu seems run), and it is not able to finish the optimization
> (but it doesn't return error).
> Changing slight primal and /or dual tolerance seems that we can go ahead,
> and in the node we obtain the following warning message
>
> bad dual - going to primal 1 524345
>
> But the same problems appear again in a further node.
>
> Normally, during the optimization, we obtain a lot of these message.
> These problem is obtained with variuos strategy and node comparing mode
>
> Can somebody help us about this problem ?
> What is a setting target to reduce "bad dual"
>
> Thank in advance
>
> Mauro
>
>
>
> Mauro SCARIONI
> IBM Global Business Services
> Circonvallazione Idroscalo
> 20090 Segrate
> Telefono 02-5962-5621
> Mobile 335-7248615
>
> IBM Italia S.p.A.
> Sede Legale: Circonvallazione Idroscalo - 20090 Segrate (MI)
> Cap. Soc. euro 384.506.359,00
> C. F. e Reg. Imprese MI 01442240030 - Partita IVA 10914660153
> Società soggetta all’attività di direzione e coordinamento di International
> Business Machines Corporation
>
> (Salvo che sia diversamente indicato sopra / Unless stated otherwise above)
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>
>



-- 
Dr. Ted Ralphs
Associate Professor, Lehigh University
(610) 628-1280
ted 'at' lehigh 'dot' edu
coral.ie.lehigh.edu/~ted





More information about the Clp mailing list