[SCIP] Does row aging depend on cut efficiacy?

[Felipe Serrano]

Dear Christian

It seems that there are two behaviors.

1. For rows in the LP. Here the rows age when they are not active. See
SCIPlpUpdateAges in lp.c

2. For cuts in the cutpool. If the cut is not in the LP, then the cuts age
is increased if its efficacy is small enough. See SCIPcutpoolSeparate in
cutpool.c

So, the efficacy plays no role for rows of the LP.

Best,
Felipe

On Sat, Nov 21, 2020 at 11:33 AM Franzen, Christian <
franzen at or.rwth-aachen.de> wrote:

> Dear SCIP team.
>
>
> Currently I am doing some experiments with row aging in a
> branch-and-price-and-cut algorithm. I am wondering if row aging is related
> to the efficiacy of a cut. My question is, when exactly row aging starts
> for a constraint a^T x <= b? So far I see two options (for current LP
> solution x*):
>
>
> 1) It starts when a^T x* - b < 0, so the cut has become obsolete
>
>
> 2) It starts when (a^T x* - b) / ||a|| < min efficiacy, so the cuts
> efficiacy has become to low
>
>
> Is any of that options right? Can you please shed some light on that
> feature implementation within SCIP. Maybe there is some documentation?
> Unfortunately, I could not find any.
>
>
> Regards
>
> Christian
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20201121/6caad37e/attachment.html>