International Journal of Computer Mathematics ( IF 1.8 ) Pub Date : 2021-06-28 , DOI: 10.1080/00207160.2021.1942459 Malihe Safarzadeh 1 , Hossein Sadeghi Goughery 1 , Abbas Salemi 2
In this paper, we present a new algorithm Global-DGMRES to find the Drazin inverse solution of the linear matrix equation AXB = C, where at least one of the matrices A or B is rank deficient. This method is based on oblique projection process, onto matrix Krylov subspaces. Also, we study convergence properties of this algorithm. The matrix equation AXB = C is a mathematical model for deblurring problems and the Global-DGMRES method helps us to reconstruct blurred images corresponding to this matrix equation. Moreover, by numerical results, we compare the proposed method with Global-LSMR and Global-LSQR.
中文翻译:
矩阵方程 AXB = C 的全局 DGMRES 方法
在本文中,我们提出了一种新算法Global-DGMRES来找到线性矩阵方程AXB = C的 Drazin 逆解,其中矩阵A或B中至少有一个是秩亏的。该方法基于斜投影过程,在矩阵 Krylov 子空间上。此外,我们研究了该算法的收敛特性。矩阵方程AXB = C是去模糊问题的数学模型,Global-DGMRES方法帮助我们重建与该矩阵方程对应的模糊图像。此外,通过数值结果,我们将所提出的方法与Global-LSMR 和 Global-LSQR 进行了比较。