A trust region filter method for general non-linear programming (Q2489398)

The authors propose two trust region filter algorithms for solving a general nonlinear programming minimization problem, that under certain conditions produces Kuhn-Tucker points. The basic idea for the algorithms is that in every step the trust region approach produces a new iteration point and some filtered rules are employed to determine whether this point is accepted by the filter set or not, the filter set being a set of points such that no point dominates any other in the sense of multi-objective terminology. These algorithms can avoid the Maratos effect.