首页> 外文学位 >Multilevel preconditioners for strongly anisotropic problems.
【24h】

Multilevel preconditioners for strongly anisotropic problems.

机译:适用于强各向异性问题的多层预处理器。

获取原文
获取原文并翻译 | 示例

摘要

In this dissertation, we develop new multilevel approaches to precondition algebraic problems stemming from the finite volume discretization of the diffusion equation with anisotropic, discontinuous coefficients. Two approaches are discussed.;In the first approach, preconditioners are based on a partitioning of the mesh in the (x, y)-plane into non-overlapping subdomains and on a special coarsening algorithm within each of the mesh layers. We show that the technique can be directly applied to the prismatic meshes based on arbitrary 2D Voronoi mesh cells. Obtained numerical results comply with the theoretical statement that the condition number of the preconditioned system receives neither an impact from diffusion tensor anisotropy nor from the thickness of thin mesh layers.;The second approach is of the algebraic multigrid type. The approach is based on the representation of the graph of the system matrix as a union of clusters. The clusters are used to design and analyze coarsening algorithms. The multilevel framework involves inner Chebyshev iterations. We construct preconditioners for two model problems (domain with a thin layer and domain with a canal) and compare performance of our preconditioners with that of another algebraic multigrid preconditioner.
机译:在本文中,我们开发了新的多级方法来求解代数问题,该方法源于具有各向异性,不连续系数的扩散方程的有限体积离散化。讨论了两种方法。在第一种方法中,预处理器基于将(x,y)平面中的网格划分为不重叠的子域,并且基于每个网格层内的特殊粗化算法。我们表明,该技术可以直接应用于基于任意2D Voronoi网格单元的棱柱形网格。所得数值结果符合理论上的陈述,即预处理系统的条件数既不受扩散张量各向异性的影响,也不受薄网格层厚度的影响。第二种方法是代数多重网格类型。该方法基于系统矩阵图作为聚类的并集表示。聚类用于设计和分析粗化算法。多层框架涉及内部Chebyshev迭代。我们针对两个模型问题(带薄层的区域和带运河的区域)构造了预处理器,并将我们的预处理器与另一个代数多重网格预处理器的性能进行了比较。

著录项

  • 作者

    Yavich, Nikolay.;

  • 作者单位

    University of Houston.;

  • 授予单位 University of Houston.;
  • 学科 Mathematics.
  • 学位 Ph.D.
  • 年度 2009
  • 页码 96 p.
  • 总页数 96
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 数学;
  • 关键词

  • 入库时间 2022-08-17 11:38:30

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号