For a large scale MINLP problem, about hundreds of thousands of variables, if I have a feasible solution, is it fast for the solver to find a local optimal solution?
If I have no feasible solution, how likely can the solver find the local optimal solution? For example,
IPOPT &
BARON.