一种求解奇异鞍点问题新的改进SSOR方法

The Study of a New Modified SSOR Method For Solving Singular Saddle Point Problems

研究了一种求解奇异鞍点问题的新的改进SSOR方法,得到其半收敛性条件及极小化拟谱半径的局部最优参数,数值例子表明选取适当的参数值可以提高算法的收敛效率.

In this paper, a new modified SSOR（NMSSOR） method for solving singular saddle point problems is studied. It is proved that the semi-convergence of the NMSSOR method under suitable restrictions on the iteration parameters is obtained. The local optimum parameters which minimize the semi-convergence and the pseudo-spectral radii of the associated iteration matrices are received. The numerical example indicates that the convergence efficiency of the NMSSOR method for solving singular saddle point problems will be improved if the appropriate parameter values are selected.