一种求解大规模稀疏二次规划的迭代法

Iterative Method for Solving Large Scale Sparse Quadratic Programming

运用部分松驰乘子法将一般二次规划化成一系列带简单约束的严格凸二次规划，然后利用广义共轭梯度法来求解这些特殊的二次规划，同时得到了原问题的对偶解。

This paper transforms a general convex quadratic programming into a series of strictly convex quadratic programming with only simple constraint by means of the method of the partial relaxation multiplier. And then,the special quadratic programming is solved by generalized conjugate gradient method, thus obtaining the dual solution to the initial problem . The convergence concerning this method and numerical results are given.