摘要
提出了一类新的求解无约束最优化问题的新拟牛顿非单调信赖域算法.采用加权的r_k用以调整信赖域半径,在适当的条件下,证明了算法的全局收敛性.数值结果表明算法的有效性.
In this paper, we propose a new quasi-Newton nonmonotone trust region algorithm for unconstrained optimization. We adjust the trust radius using not only rk, but also the previous ratios （rk-m,… ,rk}, where m is some positive integer. Under proper assumptions, we prove the global convergence of the algorithm and numerical experiments show the algorithm is competitive.
出处
《数学的实践与认识》
CSCD
北大核心
2011年第22期191-199,共9页
Mathematics in Practice and Theory
关键词
无约束最优化
新拟牛顿方程
信赖域方法
非单调线搜索
全局收敛性
Unconstrained optimization
new quasi-Newton equation
trust-region method Nonmonotonic line search
Global convergence
