<p>In this paper, a class of two-scale mixed finite element methods, including the two-scale mixed finite element method and the two-scale postprocessed mixed finite element method, is proposed for solving the Stokes problem. Both methods are grounded in the Bernardi-Raugel element. The essence of the two-scale mixed finite element solution lies in combining the mixed finite element solution from a coarse grid with those from several univariate fine grids. The underlying principle is that the low-frequency components of the mixed finite element solution can be adequately captured on the coarse grid, while the high-frequency components are more efficiently handled by the univariate fine grids. Both theoretical and numerical analyses demonstrate that the two-scale mixed finite element solution achieves the same order of accuracy as the traditional mixed finite element solution, yet at substantially lower computational costs. Additionally, we incorporate a postprocessing technique to develop and analyze the two-scale postprocessed mixed finite element method. This method is even more computationally efficient surpassing both the postprocessed mixed finite element method and the two-scale mixed finite element method.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A class of two-scale mixed finite element discretizations for the Stokes problem

  • Rong Fu,
  • Fang Liu

摘要

In this paper, a class of two-scale mixed finite element methods, including the two-scale mixed finite element method and the two-scale postprocessed mixed finite element method, is proposed for solving the Stokes problem. Both methods are grounded in the Bernardi-Raugel element. The essence of the two-scale mixed finite element solution lies in combining the mixed finite element solution from a coarse grid with those from several univariate fine grids. The underlying principle is that the low-frequency components of the mixed finite element solution can be adequately captured on the coarse grid, while the high-frequency components are more efficiently handled by the univariate fine grids. Both theoretical and numerical analyses demonstrate that the two-scale mixed finite element solution achieves the same order of accuracy as the traditional mixed finite element solution, yet at substantially lower computational costs. Additionally, we incorporate a postprocessing technique to develop and analyze the two-scale postprocessed mixed finite element method. This method is even more computationally efficient surpassing both the postprocessed mixed finite element method and the two-scale mixed finite element method.