求解加权线性最小二乘问题的一类预处理GAOR方法

A CLASS OF PRECONDITIONED GAOR METHODS FOR SOLVING WEIGHTED LINEAR LEAST-SQUARES PROBLEM

为了快速求解一类来自加权线性最小二乘问题的2×2块线性系统,本文提出一类新的预处理子用以加速GAOR方法,也就是新的预处理GAOR方法.得到了一些比较结果,这些结果表明当GAOR方法收敛时,新方法比原GAOR方法和之前的一些预处理GAOR方法有更好的收敛性.而且,数值算例也验证了新预处理子的有效性.

In this paper,a new type of preconditioners are proposed for accelerating the GAOR method,which are the preconditioned GAOR methods,for solving a class of block 2 x 2 linear systems arising from the weighted linear least-squares problem.Some comparison results are obtained,the comparison results show that the convergence rate of the proposed preconditioned GAOR methods are indeed better than those of the original GAOR method and the preconditioned GAOR methods,whenever the original GAOR method is convergent.Furthermore,effectiveness of the proposed methods is verified by numerical experiment.