Originally posted by: BCornelusse

Hello,

I would like to know what CPLEX does when the branching is of type BranchOnAny and that it creates the following child nodes
1. a node with many variables set to their LB or UB
2. a node with no variable branched on
In particular, the second child node sometimes yields an integral solution although no variable is branched on.
I suppose this is related to tolerance issues on the integrality of the solution at the parent node.
(This arises in a MIQP problem.)

Can anyone provide me some information or indicate me where I can find some?

Regards,

Bertrand.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Sorry for the late reply.

Are you saying that the second child has no changes at all, i.e., it is identical to the parent node? When you say that the child node sometimes yields an integral solution, do you mean that the relaxation is integral, or that a heuristic called at the node found an integral solution? (In the log file, heuristic solutions are marked with a '+' in the node log line.) The former case would be really strange if the child node is identical to the parent.
Tobias
#CPLEXOptimizers
#DecisionOptimization