|Title:||A self-regularized approach for rank-deficient systems in the BEM of 2D Laplace problems||Authors:||Jeng-Tzong Chen
|Keywords:||Ill-conditioned system;rank-deficient problem;self-regularized method;bordered matrix;singular value decomposition;degenerate scale;rigid body mode||Issue Date:||2017||Publisher:||Taylor & Francis Group||Journal Volume:||25||Journal Issue:||1||Start page/Pages:||89-113||Source:||Inverse Problems in Science and Engineering||Abstract:||
The Laplace problem subject to the Dirichlet or Neumann boundary condition in the direct and indirect boundary element methods (BEM) sometimes both may result in a singular or ill-conditioned system (some special situations) for the interior problem. In this paper, the direct and indirect BEMs are revisited to examine the uniqueness of the solution by introducing the Fichera’s idea and the self-regularized technique. In order to construct the complete range of the integral operator in the BEM lacking a constant term in the case of a degenerate scale, the Fichera’s method is provided by adding the constraint and a slack variable to circumvent the problem of degenerate scale. We also revisit the Fredholm alternative theorem by using the singular value decomposition (SVD) in the discrete system and explain why the direct BEM and the indirect BEM are not indeed equivalent in the solution space. According to the relation between the SVD structure and Fichera’s technique, a self-regularized method is proposed in the matrix level to deal with non-unique solutions of the Neumann and Dirichlet problems which contain rigid body mode and degenerate scale, respectively, at the same time. The singularity and proportional influence matrices of 3 by 3 are studied by using the property of the symmetric circulant matrix. Finally, several examples are demonstrated to illustrate the validity and the effectiveness of the self-regularized method.
|Appears in Collections:||河海工程學系|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.