供稿: 景书杰;徐莹莹 | 时间: 2018-11-19 | 次数: |
作者单位:河南理工大学数学与信息科学学院
摘要:将结构正割法应用到拟牛顿算法中,利用目标函数的梯度信息和函数值信息,引入拟牛顿方程,采用Wolfe线搜索准则,给出了求解无约束优化问题的一个新算法,并在一定条件下证明了新算法的收敛性和超线性收敛性。
DOI:10.16186/j.cnki.1673-9787.2015.04.028
分类号:O224
Abstract:The structure secant method is applied to the quasi-Newton algorithms,and according to utilizing the gradient information and the function value of the objective function,the introduction of quasi-Newton equation and using Wolfe line search criteria,a new algorithm for unconstrained optimization problems is given. Furthermore,local and super-linear convergence of the algorithm is obtained under some reasonable conditions.