Improving suboptimal solution for a network design problem

Hey Joris,

Thanks for the detailed description; that helps a lot. You are right that the solver should usually detect such improvements; most of our heuristics have a step in place for this. I wonder why this did not apply here. Maybe, the solution was found by a heuristic that assumed it could be bypassed (but was apparently wrong), or your model uses some special structures that lead to skipping the step (even if said variables and constraints are not part of these structures, so there would be some room for improvement). I would suggest taking this to an e-mail conversation to see if we can get to the bottom of this question.

Timo