期刊文献+

非光滑单值优化的非单调线搜索方法(英文)

A Nonmonotone Line Search Technique for Nonsmooth Unary Optimization
下载PDF
导出
摘要 提供了非光滑单值优化的非单调线搜索方法 .基于线性规划的对偶定理 ,目标函数的方向导数可以表示成线性规划问题 ,这在实际计算非单调线搜索子问题时是非常重要的 .在合理的条件下 ,证明了算法的整体收敛性和局部超线性收敛速率 . We present a nonmonotone line search algorithm for nonsmooth unary optimization problems. Based on the duality theorem of linear programming, the directional derivatives of the objective function can be expressed as a linear programming which is very important in the practical calculation for nonmonotone line search subproblems. A theoretical analysis proves that the proposed algorithm is globally convergent and has a local superlinear rate under some mild conditions.
作者 傅军
出处 《上海师范大学学报(自然科学版)》 2001年第2期34-40,共7页 Journal of Shanghai Normal University(Natural Sciences)
基金 Postgraduate Foundation of Shanghai Teachers University
关键词 非单调线搜索技术 单值优化 非线性无约束极小化 收敛性 nonmonotone line search technique unary optimization nonlinear unconstrained minimization convergence
  • 相关文献

参考文献7

  • 1GOLDFARB D,WANG S. Partial-updateNewton Methods for Unary, Factorable, and Partally Separable Optimization[J]. SIAM JOptimization, 1993, 3: 382-397.
  • 2MOCORMICK G P, SOFER A. Optimization with Unary Function[J]. MathematicalProgramming, 1991,52: 167-178.
  • 3ROCKFELLAR R T. Convex Analysis, Princeton, New Jersey[M]. Princeton UniversityPress. 1977.
  • 4ZHANG J. Superlinear Convergence of a Trust Region-Type Successive LinearProgramming Method[J]. J Op-timization Theory and Application, 1989, 61: 295-310.
  • 5GRIPPO L, LAMPARIELLO F, LUCIDI S. A Nonmonotone Line Search Technique for Newton'sMethods[J]. SIAM J. Numer Anal, 1986, 23: 707-716.
  • 6SACH E. Global Convergence of Quasi-Newton-Type Algorithms For Some NonsmoothOptimization Problems[J]. J Optimization Theory and Applications, 1983, 40: 201-219.
  • 7ZHU D T. A Trust Region Algorithm for Nonsmooth Unary Optimization[J]. Journal ofShanghai Teachers University, 1999, 28: 1-9.

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部