Lower Dimensional Coarse Spaces for Domain Decomposition
Lower Dimensional Coarse Spaces for Domain Decomposition
复制标题
用于域分解的低维粗空间
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
O. Widlund
中科院分区：
文献类型：
--
作者：
C. Dohrmann;O. Widlund
提出了一种自动构造区域分解算法中低维粗糙空间的方法。粗糙空间是完全连续的，非常适合用于两级或多级预处理器，如重叠施瓦茨。标量椭圆问题的分析表明，显着减少粗空间维数往往可以实现，而不牺牲更大的粗空间的有利条件数估计。三维的例子来说明的方法，并确认分析。
Methods are presented for automatically constructing lower dimensional coarse spaces for domain decomposition algorithms. The coarse spaces are fully continuous and well suited for use in two-level or multi-level preconditioners such as overlapping Schwarz. An analysis for scalar elliptic problems reveals that significant reductions in the coarse space dimension can often be achieved while not sacrificing the favorable condition number estimates for larger coarse spaces. Three dimensional examples are presented to illustrate the methods and to confirm the analysis.