广义拟牛顿算法对一般目标函数的收敛性  

The Convergence of Quasi-Newton Methods for Unconstrained Optimization

作　　者：王丽伟[1] 刘大莲[1] 

机构地区：[1]北京联合大学基础部,北京100101

出　　处：《北京联合大学学报》2007年第4期69-70,76,共3页Journal of Beijing Union University

摘　　要：对于无约束最优化问题minf(x),x∈Rn,提出了一种广义拟牛顿算法,并且讨论了广义拟牛顿算法对一般目标函数的全局收敛性,以及当f(x)满足Lipschitz连续的条件下,证明了相应的超线性收敛定理。A class of algorithms which are update Quasi-Newton methods for unconstrained optimization are as follows： min f（ x ）, x ∈ R^n. The proof of the global and superlinearly convergence of the generalized-quasi-Newton methords is proposed. Three main theorems are given.

关 键 词：广义拟牛顿算法 全局收敛性 超线性收敛性 

分 类 号：O242.23[理学—计算数学]