[Scip] Column generation
Gerald Gamrath
gamrath at zib.de
Thu Mar 13 08:40:07 CET 2014
Hi Mahdi,
in addition to that: When the optimal reduced cost is zero (and you
should really use some tolerance as Marco suggests!), do you add a
variable anyway? SCIP will continue pricing as long as you add variables
and only stop, if you didn't add any new variables in the last round.
Another thing: Do you add cutting planes? After that, pricing needs to
be done again.
Can you send me a log file of your run (including statistics), then I
can have a look at it.
Best,
Gerald
Am 12.03.2014 23:29, schrieb Marco Lübbecke:
>
> 2014-03-12 21:44 GMT+01:00 Mahdi Noorizadegan (DIMAP)
> <phd09mn at mail.wbs.ac.uk <mailto:phd09mn at mail.wbs.ac.uk>>:
>
> the reduced cost of column is equal to
> the main cost of a column - the dual values
>
>
> just to be sure. "main cost of a column - the dual values" means: c -
> pi * a, where c is a number (the new variable's cost), pi is the
> vector of dual variables, and a is the vector of coefficients of the
> new variable?
>
> when something happens what you describe I am almost certain that your
> reduced cost computation has a bug. Also: what is "negative"? Do you
> compare against 0.0, or do you use the SCIP methods to compare for
> negativity up to certain tolerances?
>
> Marco
>
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20140313/1ee36619/attachment.html>
More information about the Scip
mailing list