On the Local and Global Convergence of a Reduced Quasi-Newton Method

Gilbert, J.C. (1987). On the Local and Global Convergence of a Reduced Quasi-Newton Method. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-87-113

Preview

Text WP-87-113.pdf Download (1MB) | Preview

Abstract

In optimization in R^n with m nonlinear equality constraints, we study the local convergence of reduced quasi-Newton methods, in which the updated matrix is of order n-m. In particular, we give necessary and sufficient conditions for q-superlinear convergence (in one step). We introduce a device to globalize the local algorithm which consists in determining a step on an arc in order to decrease an exact penalty function. We give conditions so that asymptotically the step will be equal to one.

Actions (login required)

View Item