[SCIP] Get the branch decision on binary variables

liding xu lidingx.zz at gmail.com
Fri Jun 28 11:53:57 CEST 2019


Hello,
   I am experimenting with branch and price algorithm, and I use the
internal branch rules of scip. I want to detect which binary variable is
fixed to zero in current node. Therefore, I implemented an event handler
catching the tightening of local upper bound of variable list.
   However, after the LP solved at the root node, the event handler detects
many binary variables' local upper bounds are 0s (because of propagators?).
More confusingly after the first branch, the solver calls pricing plugin
afterwards, and only one variable is fixed to 1.
   (1).There is no branch in the root at all, why does local upper bound
change in here ?
   (2). How can the event handler catch the 'real' fixed variable of the
current node after branch?

Best,
Liding XU
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20190628/3266bd34/attachment.html>


More information about the Scip mailing list