[Scip] Lower bound in branch and price method

[Mahdi Noorizadegan (DIMAP)]

Dear SCIP,

I am trying to solve a variant of VRP in SCIP within branch and price
algorithms.
The problem is that after few iterations without any dual bound and
infinite gap, it generates a lower bound which is higher than the optimal
solution of the instance I solve.
I was wondering if you could help me to understand how SCIP generates the
dual bound and why while it is still in early stages it generates very high
lower bound?

Thanks
Mahdi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.zib.de/pipermail/scip/attachments/20130806/13c3227a/attachment.html