Real-space Kerker method for self-consistent calculation using non-orthogonal basis functions
Real-space Kerker method for self-consistent calculation using non-orthogonal basis functions
复制标题
使用非正交基函数进行自洽计算的实空间 Kerker 方法
DOI：
10.1088/0965-0393/16/3/035004
复制
发表时间：
2008
影响因子：
1.8
通讯作者：
N. Yoshikawa
中科院分区：
文献类型：
--
作者：
Y. Shiihara;O. Kuwazuru;N. Yoshikawa
我们提出了实空间 Kerker 方法，用于使用非正交基函数在实空间方法中进行快速自洽场计算。在具有许多原子的大型系统中，Kerker 方法是防止电荷晃动的一种非常有效的方法，电荷晃动会在自洽迭代过程中引起数值不稳定。我们用非正交基函数构造 Kerker 预处理矩阵，并通过求解线性方程来执行预处理。所提出的实空间Kerker方法与倒易空间中的方法相同，具有以下两个优点：（i）该方法不使用快速傅立叶变换，因此适合大规模并行计算。 (ii) 预处理是在可接受的计算时间内执行的，因为不需要执行耗时的积分，包括指数核，这与 Manninen 等人 (1975 Phys. Rev. B 12 4012) 使用的方法不同。
We have proposed the real-space Kerker method for fast self-consistent-field calculations in real-space approaches using non-orthogonal basis functions. In large-scale systems with many atoms, the Kerker method is a very efficient way to prevent charge sloshing, which induces numerical instability during the self-consistent iterations. We construct the Kerker preconditioning matrix with non-orthogonal basis functions and the preconditioning is performed by solving linear equations. The proposed real-space Kerker method is identical to the method in reciprocal space, with the following two advantages: (i) the method is suitable for massively parallel computation since it does not use the fast Fourier transform. (ii) The preconditioning is performed in an acceptable computational time since time-consuming integration, including the exponential kernel, need not be performed, unlike the method used by Manninen et al (1975 Phys. Rev. B 12 4012).