基于正则化的预处理子求解非对称广义鞍点问题

Regularized preconditioners for solving non-symmetric generalized saddle point problems

推广了文献[33]的正则化技巧用于求解非对称的广义鞍点问题.证明了相应迭代法的无条件收敛性及相应的预处理矩阵的谱性质.基于该预处理子,提出了一种松弛的预处理形式,对其预处理后的系统的特征性质给出了相关结论.通过数值试验证明了所提出的预处理子的有效性.

The regularization technique of the literature[33]was further generalized to solve non-symmetric generalized saddle point problems.The unconditional convergence property of the corresponding iterative method and the spectral properties of the corresponding preconditioned matrix were proved.Based on the preconditioners,a relaxed preconditioning form was proposed and relevant conclusions about the eigenproperties of the preconditioned system were also arrive at.The effectiveness of the proposed new precondtioners was illustrated by numerical experiments.