[Coin-discuss] Node and Cuts

Francois Margot fmargot at andrew.cmu.edu
Wed Aug 31 02:49:28 EDT 2005


> ------------------------------
>
> Message: 4
> Date: Tue, 30 Aug 2005 16:53:03 -0500
> From: "Heesu Hwang" <hxh9528 at omega.uta.edu>
> Subject: RE: [Coin-discuss] Node and Cuts
> To: "'Discussions about open source software for Operations Research'"
> 	<coin-discuss at list.coin-or.org>
> Message-ID: <000801c5adad$32529eb0$020ba8c0 at Peter>
> Content-Type: text/plain;	charset="us-ascii"
>
> Dear, Francois.
>
> One of my constrains is a quadratic constraint, and I am generating many
> Kelley cuts to linearize the constraint, one in each iteration.
> I may say they are sort of standard cutting planes.
>
> In the down branch the constraint is exactly same as what I've
> formulated.
> But in the up branch COIN inserted a slack variable, and made RHS value
> really large one.
>
> e.g., original constraint
> ... + 11099349.071 x52 <= 74721766.068
> became
> ...+ 11099349.071 x52 - Rgc31  = -1e+20
>
> Could you explain why COIN inserts the slack variable instead of using
> exactly same constraint?
>

I have never witnessed that with Bcp. My guess is that either you are doing 
something weird when generating the cuts, when passing them to Bcp,
or when you print them.

Starting with the last one, how did you get this listing? Bcp is working
internally without names, so I am a little bit surprised that you get
a slack variable with a creative name such as "Rgc31" ...

When you create the cuts (in generate_cuts_in_lp(), I assume), 
how do you pass them back to Bcp?

Francois



More information about the Coin-discuss mailing list