On the characterization of \(Q\)-superlinear convergence of quasi-Newton interior-point methods for nonlinear programming (Q1908583): Difference between revisions

It is admitted that the primal-dual Newton interior-point method can be considered as a damped and perturbed Newton method applied to the Karush-Kuhn-Tucker (KKT) necessary conditions. The KKT conditions involve the vector of primal variables \(x\), the vector of dual variables \(y\) corresponding to equality constraints and the vector of dual variables \(z\) corresponding to the nonnegative constraints on the primal variables \(x\). Introducing some assumptions on the interaction between the primal variable \(x\) and the dual variable \(z\) the authors extend the Boggs-Tolle-Wang characterization of \(Q\)-superlinear convergence for quasi-Newton methods for equality constrained optimization to quasi-Newton interior-point methods for general nonlinear programming in terms of the variables \((x, z)\) alone. Besides the determination of the primary variables, other principal problems such as the choice of the centering parameter and the choice of the steplength parameter are discussed.