[SCIP] separators and pricing
Jonas Witt
witt at or.rwth-aachen.de
Mon Sep 17 11:04:57 CEST 2018
Hi James,
If cutting planes are added to the master problem, the corresponding
dual variables need to be considered in the pricing problem. If you
don't consider these dual variables, it will make the pricing problem
invalid. Usually the structure of the pricing problem changes when cuts
that are formulated with master variables are added, i.e., additional
variables or constraints have to be added to the subproblem. How the
subproblem is changed depends on the type of the added cut. This is why
separation is disabled in the bin packing example.
Alternatively, one can add cuts from the original problem in the context
of Dantzig-Wolfe reformulation; these cuts only change the objective
function of the subproblem. But in the bin packing problem the pricing
problems are aggregated, which is why this would reintroduce a lot of
symmetry and pricing problems have to be disaggregated again.
For more details, you can have a look at the following paper:
https://doi.org/10.1002/net.20471
Best regards,
Jonas
On 09/11/2018 01:53 PM, James Cussens wrote:
> I notice that in the binpacking branch-and-price example, that all
> separators are turned off. My guess is that this is to avoid the risk
> of a separator generating an inequality that (1) becomes invalid once
> a new variable is added by the pricer and (2) where it be too complex
> to add the new variable to recover validity.
>
> Is this correct, or is there some other reason for this?
>
> Thanks,
>
> James
> --
> James Cussens
> Dept of Computer Science &
> York Centre for Complex Systems Analysis
> Room 326, The Hub, Deramore Lane Tel +44 (0)1904 325371
> University of York Fax +44 (0)1904 500159
> York YO10 5GE, UK http://www.cs.york.ac.uk/~jc
> <http://www.cs.york.ac.uk/%7Ejc>
> http://www.york.ac.uk/docs/disclaimer/email.htm
>
>
> _______________________________________________
> 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/20180917/d1c47209/attachment.html>
More information about the Scip
mailing list