A partially orthonormal basis is constructed with better conditioning properties for tetrahedral H(curl)-conforming Nedelec elements. The shape functions are classified into several categories with respect to their ...A partially orthonormal basis is constructed with better conditioning properties for tetrahedral H(curl)-conforming Nedelec elements. The shape functions are classified into several categories with respect to their topological entities on the reference 3-simplex. The basis functions in each category are constructed to achieve maximum orthogonaiity. The numerical study on the matrix conditioning shows that for the mass and quasi-stiffness matrices, and in a logarithmic scale the condition number grows linearly vs. order of approximation up to order three. For each order of approximation, the condition number of the quasi-stiffness matrix is about one order less than the corresponding one for the mass matrix. Also, up to order six of approximation the conditioning of the mass and quasi- stiffness matrices with the proposed basis is better than the corresponding one with the Ainsworth-Coyle basis Internat. J. Numer. Methods. Engrg., 58:2103-2130, 2003. except for order four with the quasi-stiffness matrix. Moreover, with the new basis the composite matrix μM + S has better conditioning than the Ainsworth-Coyle basis for a wide range of the parameter μ.展开更多
We construct and analyse a nodal O(h^4)-superconvergent FE scheme for approximating the Poisson equation with homogeneous boundary conditions in three-dimensional domains by means of piecewise trilinear functions. T...We construct and analyse a nodal O(h^4)-superconvergent FE scheme for approximating the Poisson equation with homogeneous boundary conditions in three-dimensional domains by means of piecewise trilinear functions. The scheme is based on averaging the equations that arise from FE approximations on uniform cubic, tetrahedral, and prismatic partitions. This approach presents a three-dimensional generalization of a two-dimensional averaging of linear and bilinear elements which also exhibits nodal O(h^4)-superconvergence (ultracon- vergence). The obtained superconvergence result is illustrated by two numerical examples.展开更多
文摘A partially orthonormal basis is constructed with better conditioning properties for tetrahedral H(curl)-conforming Nedelec elements. The shape functions are classified into several categories with respect to their topological entities on the reference 3-simplex. The basis functions in each category are constructed to achieve maximum orthogonaiity. The numerical study on the matrix conditioning shows that for the mass and quasi-stiffness matrices, and in a logarithmic scale the condition number grows linearly vs. order of approximation up to order three. For each order of approximation, the condition number of the quasi-stiffness matrix is about one order less than the corresponding one for the mass matrix. Also, up to order six of approximation the conditioning of the mass and quasi- stiffness matrices with the proposed basis is better than the corresponding one with the Ainsworth-Coyle basis Internat. J. Numer. Methods. Engrg., 58:2103-2130, 2003. except for order four with the quasi-stiffness matrix. Moreover, with the new basis the composite matrix μM + S has better conditioning than the Ainsworth-Coyle basis for a wide range of the parameter μ.
基金supported by Project no. 211512 from the Academy of Finlandsupported by Academy Research Fellowship no. 208628+1 种基金Project no. 124619 from the Academy of Finlandsupported by Grant IAA 100190803 of the Academy of Sciences of the Czech Republic and Institutional Research Plan AV0Z 10190503
文摘We construct and analyse a nodal O(h^4)-superconvergent FE scheme for approximating the Poisson equation with homogeneous boundary conditions in three-dimensional domains by means of piecewise trilinear functions. The scheme is based on averaging the equations that arise from FE approximations on uniform cubic, tetrahedral, and prismatic partitions. This approach presents a three-dimensional generalization of a two-dimensional averaging of linear and bilinear elements which also exhibits nodal O(h^4)-superconvergence (ultracon- vergence). The obtained superconvergence result is illustrated by two numerical examples.