摘要
针对目标函数不含交叉变量的多个可分离算子的线性约束凸优化问题,利用定制的邻近点算法,线性化算法迭代的二次项,将其转变为单调的变分不等式子问题,给出一种新的线性化定制的邻近点算法.结果表明:对于多个可分离的线性约束凸优化问题,线性化定制的邻近点新算法是有效的,将其转化为等价的混合变分不等式形式,证明了算法的全局收敛性及解的唯一性.
For the convex minimization problem with linear constraints and a block separable objective function which is represented as f＇mite dimension functions without coupled variables, based on the customized proximal point algorithm, this study linearizes the second item of the iteration subproblem, converts it to monotone problem, and offers the new algorithm. The results show that there exists the linearized method for the above problem which has global convergence and the unique solution, using its corresponding mixed variational inequality.
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2014年第7期992-995,共4页
Journal of Liaoning Technical University (Natural Science)
基金
国家自然科学青年基金资助项目(G010303)
辽宁省教育厅基金资助项目(L2012105)
关键词
变分不等式
定制邻近点算法
全局收敛性
交替方向法
矩阵范数
预测-校正迭代
凸函数
线性化算法
variational inequality
customized proximal point method
global convergence
alternating direction method
matrix norm
prediction-correction iternation
convex function
linearized method
