摘要
利用ε有效广义约束向量所张成的向量锥方法,通过求解一个低维的简单线性规划子问题获得一个下降可行方向,从而构造出求解某类二次规划的算法,并对算法进行了收敛性分析与数值计算,从算法的构造来看,它适用于大规模问题的求解。
Using the vector cone spanned by the vectors corresponding to the generalized active ε constraint and solving a lower dimensional simple LP subproblem to obtain a feasible descent direction, we construct an algorithm for solving a class of quadratic programming problems. We also give its convergence analysis and results of numerical experiments which are satisfying. In the aspect of algorithm construction, this algorithm is suitable for solving large scale problems.
关键词
二次规划
ε容许方向法
线性规划
Quadratic programming, Generalized active ε constraints, LP subproblem, Feasible descent direction