Global optimization of nonlinear bilevel programming problems (Q5942317)

The known approaches to bilevel programming problems (BLPP) normally reduce BLPP to an one level programming problem replacing the lower level minimization problem with equations of KKT conditions. However, strong assumptions should be satisfied in order to guarantee the solution to the last problem be also the solution to the original problem. The proposed approach, although uses the KKT conditions, can overcome the limitations of the previous methods. A developed method is based on the branch and bound framework known as \( \alpha\)BB. For the problems involving twice differentiable nonlinear nonconvex functions epsilon global optimality is theoretically guaranteed in finite number of iterations. Several known examples of BLPP problems are solved using the proposed method.