带有线性约束的超定线性方程组的L_1-模逼近算法

An Algorithm for Overdetermined Linear System L_1 Approximation with Linear Constraints

提出一个求解线性约束的离散化L_1-模逼近问题的新算法,该算法的理论基础是求解线性规划的有效集法,因此问题的维数不会增大,数值例子表明,该算法是非常有效的。

We propose an algorithm, with based on active set method of linear programming, for solving discrete L1 approximation problem with any type of linear constraints. The dimension of varibles in the problem is no longer extend when the problem is solved by the algorithm which is propose in this paper. The numerical result reported here indicate that the algorithm is very efficient.