A new self scaling PCG algorithm with dynamical retards

Abstract The problem of optimizing a certain systems as a function of time may be formulated in terms of maximizing or minimizing a non-linear cost function. It was found that a Dynamic Programming (DP) algorithm was particularly efficient procedure for solving this problem compared with an alternative nonlinear programming method. In this work, a new CG method with dynamical retards is generalized and combined in a dynamical way with non-monotone globalization strategies to obtain a new type CG-algorithm for minimizing non-quadratic functions that can deal efficiently with large scale nonlinear optimization problems.