Solving nonconvex nonlinear programming problems via a new aggregate constraint homotopy method

摘要

In this paper, by introducing C~2 mappings ξ _i(x_iz_i),i = 1,...,m and using the idea of the aggregate function method, a new aggregate constraint homotopy method is proposed to solve the KarushKuhnTucker (KKT) point of nonconvex nonlinear programming problems. Compared with the previous results, the choice scope of initial points is greatly enlarged, so use of the new aggregate constraint homotopy method may improve the computational efficiency of reduced predictorcorrector algorithms.