Abstract:After a quick review of the domain decomposition methods, and, more particularly on their application to large size problems relative to radiation and scattering of time-harmonic waves, we describe two contributions of the authors in this context. The two contributions are related to the scattering of a time-harmonic electromagnetic wave by a large perfectly conducting structure including a deep cavity. The first contribution is a substructuring technique. It is used to increase the speed of the convergence of… Show more
“…The Schwarz' preconditioner, as formulated in (11), has been used in earlier works for the computation of the Helmholtz problem in bounded domains (see [29,30]) and for the scattering waves (see [10,13,14]). …”
Section: Iterative Methods With Schwarz' Preconditionermentioning
SUMMARYWe consider the Jacobi preconditioner of the GMRES method introduced by Liu and Jin for the scattering problem (IEEE Trans. Ante. Prop. 2002; 50:132-140). We explain why it is a particular form of the Schwarz' preconditioner with a complete overlap and specific transmission conditions. So far, a superlinear convergence has been predicted by the general theory without any additional indication on the convergence rates. Here, we establish error bounds that provide accurate convergence rates in two and three dimensions. Courant-Weyl's min-max principle applied to some kernel operators together with some polynomial approximation estimates are the milestones for the proofs.
“…The Schwarz' preconditioner, as formulated in (11), has been used in earlier works for the computation of the Helmholtz problem in bounded domains (see [29,30]) and for the scattering waves (see [10,13,14]). …”
Section: Iterative Methods With Schwarz' Preconditionermentioning
SUMMARYWe consider the Jacobi preconditioner of the GMRES method introduced by Liu and Jin for the scattering problem (IEEE Trans. Ante. Prop. 2002; 50:132-140). We explain why it is a particular form of the Schwarz' preconditioner with a complete overlap and specific transmission conditions. So far, a superlinear convergence has been predicted by the general theory without any additional indication on the convergence rates. Here, we establish error bounds that provide accurate convergence rates in two and three dimensions. Courant-Weyl's min-max principle applied to some kernel operators together with some polynomial approximation estimates are the milestones for the proofs.
“…Also, the solution of the BEM can be performed using an adapted non-overlapping DDM (see [32]) so that each iteration is carried out by mainly doing one iteration for the FEM and another one for the BEM. In this manner, only small-size systems have to be solved at each iteration.…”
SUMMARYNon-overlapping domain decomposition techniques are used to solve the finite element equations and to couple them with a boundary element method. A suitable approach dealing with finite element nodes common to more than two subdomains, the so-called cross-points, endows the method with the following advantages. It yields a robust and efficient procedure to solve the equations resulting from the discretization process. Only small size finite element linear systems and a dense linear system related to a simple boundary integral equation are solved at each iteration and each of them can be solved in a stable way. We also show how to choose the parameter defining the augmented local matrices in order to improve the convergence. Several numerical simulations in 2D and 3D validating the treatment of the cross-points and illustrating the strategy to accelerate the iterative procedure are presented.
“…Many references are available in the literature [1][2][3][4][5][6][7][8]. They can be used as a process of distributing data from a computational domain among the machines of a cluster of PCs, or for the separation of the physical domain into regions which occurs when using different and adapted solving methods, and finally as a process of subdividing the solution of large linear system into smaller problems whose solutions can be used as a pre-conditioner or a solver.…”
Section: Domain Decomposition Methods Applied To Radiation and Scattementioning
confidence: 99%
“…These functions f n and g n are extended by zero on all of the other surfaces of the domain. Finally the electromagnetic field decomposition (11) is introduced in relation (9) and then in relation (8), yielding to the following decomposition of the currents on the equipment interface: 9) is the electric current induced on Γ by the nth magnetic current K n = f n × n belonging to Γ and extended by zero on the other surfaces of the scatterer. In the same way, J source solution of Eq.…”
Section: Scattering Matrix Of the Unbounded Volume Vmentioning
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.