首页 | 官方网站   微博 | 高级检索  
     

一类不可微规划算法的线性收敛性
引用本文:郑颖春,王雪峰.一类不可微规划算法的线性收敛性[J].西北轻工业学院学报,2013(3):154-156,170.
作者姓名:郑颖春  王雪峰
作者单位:西安科技大学理学院
基金项目:国家自然科学基金项目(71103143)
摘    要:介绍了一类不可微优化的次梯度算法,并结合文献9],给出了次梯度算法的简化形式;通过引进二阶方向导数的概念,证明了不可微函数的一阶及二阶积分中值定理,这些中值定理和可微函数的中值定理是类似的;利用这些中值定理,借助文献9]的证明方法,证明了一般的次梯度算法都具有线性收敛性.

关 键 词:不可微规划  算法  线性收敛性

The linear convergence nondifferentiable of an algorithm for programming
ZHENG Ying-chun,WANG Xue-feng.The linear convergence nondifferentiable of an algorithm for programming[J].Journal of Northwest University of Light Industry,2013(3):154-156,170.
Authors:ZHENG Ying-chun  WANG Xue-feng
Affiliation:(College of science,Xi′an University of Science & Technology,Xi′an 710054,China)
Abstract:This paper introduces a kind of sub-gradient algorithm, and combining with docu- ment9],given simplified form of the sub-gradient algorithm by introducing the concept of second order directional derivative, proved first and second order integral mean value theo- rem of the nondiffenrentiable function,The mean value theorem and the mean value theorem of differentiable functions is similar; by the mean value theorem,with the help of the method of document9],we proof that the general sub-gradient algorithm has linear convergence.
Keywords:nondifferentiable optimization aIgorithm linear convergence
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号