摘要
对于具有非线性等式约束且变量有界的非线性规划问题,提出了一个由三阶段组成的广度既约梯度变位算法,即线性近似、既约梯度求极小和可行变位阶段.同时我们证明了该算法所具有的收敛性.
We present a new algarithm for soling nonlinear programming subject to nonlinear equality constraints with simple bounds in this paper,by use of the generalized reduced gradient restortion and feasible restoration.Simutaneously we prove a large scope convergence of the algorithm.Morever the initial point is not necessary feasible.but it satiffies bounded constaints in our algorithm.
出处
《新疆大学学报(自然科学版)》
CAS
1998年第4期9-14,共6页
Journal of Xinjiang University(Natural Science Edition)
关键词
广义既约梯度
非线性规划
变位算法
收敛性
nonlinear equality constraints\ Generalized reduced gradient\ Feasible restoration method\ Kuhn-Tucker point