求解凸二次规划的一种改进的原-对偶内点算法  

Improved Primal-dual Feasible Interior Point Algorithm for Convex Quadratic Programming

作　　者：杨春艳[1] 雍龙泉[2] 

机构地区：[1]银川大学数学系,宁夏银川750105 [2]陕西理工学院数学系,陕西汉中723001

出　　处：《长江大学学报（自科版）（上旬）》2009年第2期126-128,共3页JOURNAL OF YANGTZE UNIVERSITY (NATURAL SCIENCE EDITION) SCI ＆ ENG

摘　　要：基于牛顿方向,给出了求解凸二次规划问题的改进原对偶可行内点算法。若获得算法的初始可行内点,则该算法经过多次迭代之后收敛到原问题的一个最优解。数值试验表明了该算法的有效性。In this paper we analyzed the most common algorithms to quadratic programming and indicated difficulty in studying this problem.Based on above,we presented an improved primal-dual feasible interior point algorithm for convex quadratic programming by means of the Newton direction.It is showed that if a strictly feasible starting point is available,then the algorithms have the polynomial complexity.Numerical results are demonstrated very good computational performance on convex quadratic programming.

关 键 词：凸二次规划 原对偶可行内点算法 多项式复杂性 

分 类 号：O224[理学—运筹学与控制论]