Domain decomposition and preconditioning techniques in boundary element methods
boundary element methodsboundary integral equationsCalderon projectiondomain decomposition methodsfast algorithmsGalerkin discretizationhypersingular integral operatoriterative methodsmultigrid schemesnumerical examplespreconditioningSchur complementspectral condition numberSteklov-Poincaré operator
Boundary value problems for second-order elliptic equations (35J25) Iterative numerical methods for linear systems (65F10) Numerical computation of matrix norms, conditioning, scaling (65F35) Boundary element methods for boundary value problems involving PDEs (65N38) Multigrid methods; domain decomposition for boundary value problems involving PDEs (65N55)
Domain decomposition methods are powerful tools to parallelize the implementation of boundary element methods, especially if different material parameters in substructures appear. Boundary element methods are based on equivalent boundary integral equations of corresponding mixed boundary value problems. We restrict ourselves to the direct approach for selfadjoint problems of second order, where the mapping properties of the resulting boundary integral operators are well known. The basic tool for the boundary element domain decomposition formulation is the Calderon projection in terms of two boundary integral equations from which we can derive two representations of the Dirichlet-Neumann map based on the so-called Steklov-Poincaré operator. According to the choice of these operators, we distinguish between the symmetric and the nonsymmetric formulation. Here we give a general approach for the preconditioning of discrete boundary integral operators, which is well suited for the symmetric formulation and which has no restrictions to the space dimension and the discretization scheme. The construction of global preconditioners for the discrete Steklov-Poincaré operator in its symmetric representation is based on the spetral equivalence to the Galerkin discretization of the hypersingular integral operator. This allows us to construct several preconditioners of hierarchical type, where the spectral condition number depends on the relation between the local and the diameter of the substructures. Due to the spectral equivalence described above, we may also take the discrete hypersingular integral operator as an optimal preconditioner of the global Schur complement system. The realization of this preconditioner, i.e. the inversion of the discrete hypersingular operator can be obtained either by multigrid schemes or any other preconditioned iterative methods. Based on these preconditioners we construct fast algorithms for the iterative solution of linear algebraic systems resulting from mixed boundary value problems in a variational formulation with respect to a given domain decomposition. Numerical examples demonstrate the efficiency of the proposed methods. For the entire collection see [Zbl 0877.00013].
- Symmetric boundary element method versus finite element method
- Preprocessing for 2D FE-BE domain decomposition methods
- Boundary element tearing and interconnecting domain decomposition methods
- Challenges and Applications of Boundary Element Domain Decomposition Methods
- scientific article; zbMATH DE number 4076460 (Why is no real title available?)
- scientific article; zbMATH DE number 1070422 (Why is no real title available?)
- scientific article; zbMATH DE number 218264 (Why is no real title available?)
- scientific article; zbMATH DE number 839239 (Why is no real title available?)
- Preconditioning for boundary element methods in domain decomposition
- A domain decomposition method based on BEM and FEM for linear exterior boundary value problems
This page was built for publication: Domain decomposition and preconditioning techniques in boundary element methods
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4371009)
