Hello everyone,
I am experiencing issues with my model in GAMS. I am using CONOPT to solve my model and use its solution as an initial guess for Baron to find the global optimum solution. My problem is that Baron does not assume the solution from CONOPT is feasible, and I am unsure why. Has anyone experienced that? Is there a way to check why the CONOPT solution is not feasible for BARON?
P.S.: The problem is an NLP.
Hi,
You could turn on the full equation listing by setting limrow to some large number and then check whether some equations are flagged INFES. That would mean that the starting point isn’t feasible.
An alternative could be to fix the starting point via variable attribute .fx and then run Baron with option compIIS to let Baron compute an Irreducible Inconsistent Set.
You could also share your code here to let other forum users reproduce the observed behavior which may increase the chance to get good suggestions on how to resolve the issue.
I hope this helps!
Fred