摘要
本文将梯度投影与拟牛顿法相结合，给出了求解一般线性约束非线性规划问题含两组参数的算法族．在一定的条件下证明了算法族的全局收敛性与它的子族的超线性收敛速度，并给出了投影Ｄ．Ｆ．Ｐ方法、投影ＢＦＧＳ方法等一些特例．
in this paper, we will give a class of algorithms containning two group parameters for noallnear progranuliing problems with linear constraints by combining the gradient projection with quasi-Newton method. Its global convergence and the superlinear convergence of its some subclass are proved under suitable conditions, and some special examples of the algorithm are given, such as DFP projection method, BFGS projection method and so on.
出处
《系统科学与数学》
CSCD
北大核心
1996年第2期105-112,共8页
Journal of Systems Science and Mathematical Sciences
基金
国家自然科学基金
广西自然科学基金
广西教委科学基金
关键词
非线性规划
梯度投影
拟牛顿法
超线性收敛
Nonlinear programming
gradient projection
quasi-Newton method, global and superlinearly convergence
