A MIXED AND SUPERLINEARLY CONVERGENT ALGORITHM FOR CONSTRAINED OPTIMIZATION.

In this paper, a mixed QP-free method for constrained optimization problem is proposed. This new method consists of two different sub-methods, sub-method I and sub-method II. These two sub-methods are all QP-free. A switch is designed to decide which sub-method should be adopted in each iteration. Under some mild assumptions, we prove that the method is globally and superlinearly convergent. [ABSTRACT FROM AUTHOR]