[SCIP] Variable bounds after using SCIPfixVar

[Benjamin Müller]

Hi Oliver,

I could imagine that you have observed a fixing in a sub-SCIP and thus 
the corresponding variable in your main SCIP is still unfixed. Can you 
add an artificial loop to the REDCOST callback of your pricer and check 
the bounds and the LP solution value of each variable? Do you still see 
a variable that has an LP solution value that is not in its bounds?

Best,
Benny

Am 24.10.19 um 22:42 schrieb Oliver Zier:
> Hi all,
>
> I am implementing branch and price using SCIP to solve a scheduling
> problem. I used the binpacking problem as my template and my conshdlr is
> pretty the same as cons_samediff. However, I ran into problem with fixing
> variables e.g there are 5 columns and 4 are fixed to 0 (sum of these 5
> variables must be 1) so the remaining variable is sure to be 1 but
> master's solution gives one of the fixed-to-0 variables value 1. So my
> master solution does not change at all after branching. I checked the
> local LB and UB of variables after SCIPfixVar and just after it enters
> scip_redcost and the bounds look correct to me. Can anyone give me ideas
> what can go wrong here / what I can check?
> Thank you in advance.
> Best,
> Oliver
>
>
>
>
>
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip