An active set modified Polak-Ribiére-Polyak method for large-scale nonlinear bound constrained optimization (Q1935275)

The authors propose a modification of the Polak-Ribiére-Polyak method for large-scale nonlinear optimization problems with box constraints. The modification results from an alternative approach which, at each step of the iteration, needs less storage to update both the active and non-active variables. Sufficient conditions are given (including Lipschitz continuity) guaranteeing global convergence. Numerical examples are presented demonstrating the applicability of the algorithm.