Hello. I’m solving a MINLP minimization problem with Baron and it gives me a result. I save the results with the command “model.savepoint=1;” and run them as a starting point to solve the same model with DICOPT (execute_loadpoint name.gdx’; ). In this situation, the model gives me a MUCH higher result even though the optca it’s equal to zero. This is not the first time this has happened to me. How can I fix the initial point to verify that with both solver it finds the same result?
thanks!!!
BARON can prove global optimality for non-convex models. DICOPT guarantees global optimality only for convex models and is a heuristic for non-convex problem which provides you hopefully with a good solution. It seems that your model falls into the non-convex category, hence you can’t get what you want.
-Michael