[SCIP] Variable Upper Bound is ignored

[Norbert Rümelin]

Dear Mailinglist,

I'm trying to implement a model for my Master Thesis in SCIP.
Therefor I wrote began with the TSP-Example an it seemed working until i
also tried to implement an Branching-Rule.

It seems like the SCIPchgVarUbNode has no effect.

I have written a Ryan and Foster-like Branchingrule. There I set some
Variables of an "independent" part of the problem to 0 for a
"same"-node, and the rest to 0 for a "differ"-node.
Within the pricing I get, that this variables really have the upper
bound set to 0 (and lower bound also 0 since I create them as Binary
with lb=0 and ub=1) and also that there is no creation of the same
variable again.
But after one solution-iteration I come again into the branching where
the same Variables have the same fractional solutions, as if nothing
would have happend.
It looks like the problem gets infeasible, but instead of letting the
pricer performe farkas-pricing scip decides to ignore the upper-bound I set.

Thanks for your help and sorry for my bad English
Norbert