框式凸二次规划原始-对偶势下降内点算法

A Primal-dual Interior-point Algorithm Based on Reduced Potential Function for Convex Quadratic Programming with Box Constrains

基于线性规划原始-对偶内点算法的思想,对框式凸二次规划提出了一种新的内点算法—原始-对偶势下降内点算法.算法取牛顿方向作为迭代方向,利用势函数选择迭代步长,并证明了新算法具有O(nL)的迭代复杂性.

A new algorithm, primal-dual potential-reduction interior-point algorithm, is devised. This new algorithm is based on the ideas of the primal-dual interior-point algorithm for linear programming. This algorithm uses the Newton-direction as its iteration direction; its iteration step is determined by potential func- tion; and the polynomial iteration complexity of O（nL） iterations of this new algorithm is proved.