Projected Hessian updating algorithms for nonlinearly constrained optimization. (English) Zbl 0593.65043

This paper is concerned with quasi-Newton methods for minimizing a function subject to equality constraints. Specifically, an approximation to the Hessian of the Lagrangian function is considered. The methods differ in which part of this matrix is subject to updating. Proof of two- step Q-superlinear convergence is given for a ”two-sided projected Hessian” updating algorithm, as well as numerical comparisons with other methods. All the methods are used without line-searches, but this may have a decisive influence on the performance.
Reviewer: H.Matthies


65K05 Numerical mathematical programming methods
90C30 Nonlinear programming
Full Text: DOI