ABSTRACT
There are three optimality conditions for a QP: primal feasibility, complementary slackness and dual feasibility. In Chapter 5, we developed QP solution algorithms which maintained the first two conditions at each iteration and the sequence of primal objective values is decreasing. Only upon termination is the third optimality (dual feasibility) satisfied. In this chapter, we develop a solution method for a QP which generates a sequence of points satisfying dual feasibility and complementary slackness at each iteration. The corresponding sequence of primal objective function values is increasing. All but the final point is primal infeasible.
