[SCIP] Running CG again without branching

[Gerald Gamrath]

Dear Lan,

I don't think it is intended behavior that you create new variables in 
the branching callback. You may try to do it though and add them via 
SCIPaddVar(). What you can definitely do is fix some of the variables to 
0. In that case, you should set the result to SCIP_REDUCEDDOM. Then, 
SCIP will reoptimize the LP of the current node afterwards (including 
price-and-cut loop) and call branching again, if still needed.

Best,
Gerald

On 29.10.19 21:38, Ngô Lan wrote:
>
> Hi all,
>
>
> I am implementing BPC using SCIP. Is it possible that during 
> branching, I do not create any branch but only add variables and then 
> start column generation again? My problem is that during my branching, 
> it could be possible that there are still fractional variables but no 
> other possible branching condition is found, then what I would like to 
> do is for example fix the current fractional variables to 0 and then 
> add a new variable that is a combination of those fractional 
> variables. After that instead of creating new child nodes, I would 
> like to run column generation. However, I don't know how to signal 
> scip to do that. I have tried to return branching with result 
> SCIP_DIDNOTRUN but then SCIP will continue with variable branching.
>
> Thank you for your time.
>
>
> Best,
> Lan
>
>
> _______________________________________________
> 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/20191030/cfe9e6a8/attachment.html>