[SCIP] Degeneracy and abs function

[Jennifer Uebbing]

Dear SCIP team, 

I am using SCIP to solve an MINLP optimization problem and I have a few question, which I hope you can help me with: 

- What influence does degeneracy of the MINLP (A point that does not satisfy LICQ) have? I imagine, that feasible, degenerate points can be falsely identified as a local optimum in subproblems by NLP solvers. This would give poor upper bounds and branching would take more time. Are there any other consequences? Does it effect the lower bound? Can it cause SCIP to give false results? 

- Does SCIP call NLP solvers after all integer variables have been fixed? Or are NLP solvers already called before that? 

- How does SCIP handle the abs function? In which form is it given to the solver? 

I see forward to your answers. Thank you for your time. 

Best wishes, 
Jennifer 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20210120/4859735c/attachment.html>