A new, efficient, and highly accurate numerical method which achieves the residual reduction with the aid of residual equations and the method of least squares is proposed for boundary value problems of elliptic partial differential equations. Neumann, Dirichlet, and mixed boundary value problems of
Multilevel Minimal Residual Methods for Nonsymmetric Elliptic Problems
β Scribed by Gerhard Starke
- Publisher
- John Wiley and Sons
- Year
- 1996
- Tongue
- English
- Weight
- 756 KB
- Volume
- 3
- Category
- Article
- ISSN
- 1070-5325
No coin nor oath required. For personal study only.
β¦ Synopsis
The subject of this paper is to study the performance of multilevel preconditioning for nonsymmetric elliptic boundary value problems. In particular, a minimal residual method with respect to an appropriately scaled norm, measuring the size of the residual projections on all levels, is studied. This norm, induced by the multilevel splitting, is also the basis for a proper stopping criterion. Our analysis shows that the convergence rate of this minimal residual method using the multilevel preconditioner by Bramble, Pasciak and Xu is bounded independently of the mesh-size. However, the convergence rate deteriorates with increasing size of the skew-symmetric part. Our numerical results show that by incorporating this into a multilevel cycle starting on the coarsest level, one can save fine-level-iterations and, therefore, computational work.
π SIMILAR VOLUMES
## Abstract In this paper, we review the development of local discontinuous Galerkin methods for elliptic problems. We explain the derivation of these methods and present the corresponding error estimates; we also mention how to couple them with standard conforming finite element methods. Numerical
## Abstract In this paper we use the method of layer potentials to study __L__^2^ boundary value problems in a bounded Lipschitz domain Ξ© for a family of secondβorder elliptic systems with rapidly oscillating periodic coefficients. Defining ${\cal L}\_\varepsilon = - {\rm div}(A(\varepsilon ^{ - 1}
We consider a numerical enclosure method with guaranteed L Ο± error bounds for the solution of nonlinear elliptic problems of second order. By using an a posteriori error estimate for the approximate solution of the problem with a higher order C 0 -finite element, it is shown that we can obtain the g