[Scip] How to update separators in a branch-price-and-cut?

nikolaj@crt.umontreal.ca nikolaj at crt.umontreal.ca
Wed Aug 25 16:24:33 MEST 2010


On Tue, August 24, 2010 11:41 am, Marco Luebbecke wrote:
> gamrath> You are right. Cutting plane separation complicates the gamrath>
> column generation process in most cases.
>
> Actually, the contrary is true. Cutting planes (at least those which
> are in the current base of knowledge in B&P&C) have a complicating impact
> on the pricing only in a few cases. You "only" have to know how to do
> this, that is, which types of cutting planes to separate. Try picking up
> something on "cutting planes on the original variables" (that is the same
> theme as "branching on original variables"). "Original" variables are
> those variables which are still present in the pricing problem (the
> variables of the model on which DW Dantzig-Wolfe decomposition was applied
> to obtain your master problem). When you formulate cuts in these
> variables, and Dantzig-Wolfe reformulate them to add them to the master
> problem, the only thing that changes in the pricing is its objective
> function.
>
> Best,
> Marco

Hello Marco and everybody else,

Thank you very much for your comment. I saw your article
"Branch-Price-and-Cut Algorithms" (and I'll stick to your terminology). I
will read it carefully in the coming days.

I would like to have a total control on some of the SCIP_ROWs I add/remove
to/from the LP. These rows should be updated in the pricing process. Here
are some questions.

If I use SCIPaddCut with the flag forcecut set to TRUE, is it guaranteed
that SCIP will add the SCIP_ROW? Or will it perform some (hidden) tests
before and only add the cut if these tests are successfully passed?

If I don't want SCIP to remove the SCIP_ROW I added, I guess that I need
to set the flag removable in SCIPcreateRow to FALSE. Is this enough to
have a guarantee the SCIP_ROW will stay in the LP?

How do I do to delete from the LP (and the cutpool) some of the added
SCIP_ROWs (I don't touch the initial model contraints)?

The LPs I have to solve are numerically very unstable and I need to be
able to add some cuts even if they are "poor". At the same time, the LPs
are huge, so I need to be able to reduce them as much as possible as soon
as possible.

Thank you very much in advance.

Have a nice day/night,

Nikolaj





More information about the Scip mailing list