No label defined (Q4239076)

Preface: The solution of elliptic boundary value problems may have anisotropic behaviour in parts of the domain. That means that the solution varies significantly only in certain directions. Examples include diffusion problems in domains with edges and singularly perturbed convection-diffusion-reaction problems where boundary or interior layers appear. In such cases it is an obvious idea to reflect this anisotropy in the discretization by using anisotropic meshes with a small mesh size in the direction of the rapid variation of the solution and a larger mesh size in the perpendicular direction. Anisotropic meshes can also be advantageous if surfaces with strongly anisotropic curvature (the front side of a wing of an airplane) or thin layers of different material are to be discretized. In order to describe anisotropic elements mathematically we introduce the term aspect ratio. The aspect ratio is the ratio of the diameter of the element \(e\) and the supremum of the diameters of all balls contained in \(e\). A finite element is called anisotropic if the aspect ratio tends to infinity when the mesh size or some (small perturbation) parameter tends to zero. Contrary, elements are called isotropic if the aspect ratio is bounded by a moderate constant. Triangular elements are isotropic if they satisfy Zlámal's minimal angle condition. Already in the fifties and seventies it was shown that certain local interpolation error estimates can be proved for some classes of anisotropic finite elements. The minimal angle condition is replaced by the weaker maximal angle condition. Nevertheless, the majority of papers and books on the finite element method excludes anisotropic finite elements. Since the end of the eighties anisotropic elements are considered in the international literature more intensively. Examples of using anisotropic elements include interpolation tasks, singular perturbation and flow problems, the treatment of edge singularities, and adaptive procedures. The corresponding papers lead to two conclusions. First, anisotropic mesh refinement offers a great potential for the construction of efficient numerical procedures (interpolation, finite element method, boundary element method, finite volume method), more efficient than it is possible with the restriction to a bounded aspect ratio. So one can expect a broad utilization of such meshes. Second, there are still challenges to set all the ingredients of such methods on a solid mathematical basis. These ingredients include a-priori and a-posteriori error estimates and the solution of the arising system of algebraic equations. The aim of this monograph is to establish interpolation and approximation properties of finite element spaces on anisotropic meshes. In particular, such topics are chosen where the author himself contributed to the development: anisotropic local interpolation error estimates for several types of two- and three-dimensional finite elements and a-priori estimates of the discretization error for model problems with edge singularities or boundary layers. Several results have not been presented before. We are restricted here to model problems since detailed knowledge of properties of the solution is necessary. However, much effort is spent to treat arbitrary polygonal/polyhedral domains and finite elements of any approximation order. It is a future task to apply these results to more complex problems and to complement them with mathematically founded adaptive strategies and optimal preconditioning techniques for solving the arising systems of linear equations.