Nonsmooth equation based BFGS method for solving KKT systems in mathematical programming (Q5942348): Difference between revisions

The authors propose a Broyder-Fletcher-Goldfarb-Shanno method for solving the Karush-Kuhn-Tucker system of a constrained optimization problem with finitely many variables, equality and inequality constraints, where all functions are assumed to be twice continuously differentiable. The method is based on a nonsmooth equation reformulation of the KKT system in terms of the Fisher-Burmeister function. Under some conditions the method is well defined and converges globally and superlinearly to a KKT point without any convexity assumptions on the problem. This paper is clearly written and ends with a useful discussion of the method. From the paper it is not clear how restrictive the conditions are that lead to global or superlinear convergence. Imbedding of this paper in the existing research on this subject is adequate, and the paper ends with an extensive list of references.