摘要
本文对非线性规划问题给出了一个具有一步超线性收敛速度的可行方法，由于此算法每步迭代均在可行域内进行，并且每步迭代只需计算一个二次子规划和一个逆矩阵，因而算法具有较好的实用价值，本文还在较弱的条件下证明了算法的全局收敛和一步超线性收敛性。
In this paper, a one-step superlinearly convergent feasible algorithm for nonlinear programming with nonlinear onstraints is proposed. Since the point generated by the algorithm is feasilbe per iteration and the algorithm only needs solve a sub-quadratic programming and a inverse matrix, the new algorithm has some good properties for tractical use. Under some milder conditions, the global convergence and local superlinear convergence are proven in this paper.
出处
《应用数学学报》
CSCD
北大核心
1995年第4期579-590,共12页
Acta Mathematicae Applicatae Sinica
基金
国家自然科学基金
