Hi Everyone,
I solved my nlp model using BARON and got two different solutions, feasible and the best possible, please refer to below screenshot:
Among these two solutions, which one of them is the Global Optimal Solution?
Thanks
zar
Branch and bound methods (as implemented in Baron) work with a relaxation (“Best possible”) and an incumbent (“Solution”). If the difference (absolute (“Absolute gap”) and relative (“Relative gap”)) between the two is 0 then your incumbent is the optimal solution. Since you have the tolerance for the relative gap at 0.1 Baron is allowed to terminate earlier and you won’t get the global optimum. Use “option optcr=0;” in your GAMS model to request the global solution. Moreover, it seems that Baron ran out of time (your relative gap does not fulfill the tolerance requirement), you probably have to give it more time. B&B methods have exponential running time (worst case), so you might not be able to solve your model to global optimality in reasonable time.
-Michael
Thank you very much Michael.