线性规划鞍点算法原理与实际计算(英文)  

Principle and Computational Experience of a Saddle Point Algorithm for Linear Programming

作　　者：尚毅[1] 成孟金[1] 张国光[1] 李鹏辉[1] 刘冬[1] 

机构地区：[1]沈阳化工学院,辽宁沈阳110142

出　　处：《沈阳化工学院学报》2004年第2期138-143,共6页Journal of Shenyang Institute of Chemical Technolgy

摘　　要：　阐述了线性规划鞍点算法原理与实际计算.讨论了在基不变条件下,算法的收敛性和迭代步长的选择问题.指出算法性质为q 线性收敛.鞍点算法软件与美国优化技术中心编制的内点算法软件PCx进行了比较,用两种算法计算网上NETLEB的线性规划问题,公布了比较结果.线性规划鞍点算法已应用到石化企业的优化生产工作.This paper proposes a principle of new algorithm for linear programming based on saddle point. The convergence and choice of the step size under the condition of the changeless base set is discussed. The computational complexity is q-linear. The NETLIB problems were computed by using the proposed algorithm and the interior-point working code PCx respectively, and results were compared. This code has been successfully used to solve problems in petrochemical industry.

关 键 词：线性规划鞍点算法 计算方法 并行算法 梯度 收敛性 迭代步长 化工行业 

分 类 号：TQ015.9[化学工程]