Abstract:SUMMARYOptimized Schwarz methods are working like classical Schwarz methods, but they are exchanging physically more valuable information between subdomains and hence have better convergence behavior. The new transmission conditions include also derivative information, not just function values, and optimized Schwarz methods can be used without overlap. In this paper, we present a new optimized Schwarz method without overlap in the 2D case, which uses a different Robin condition for neighboring subdomains at th… Show more
“…Similar homographic best approximation problems also occur in the design of optimized Schwarz methods for steady problems, and so far these problems have always been treated by direct analysis; see for example [26,28,27,9] for advection diffusion problems, [7,6,19,16] for indefinite Helmholtz problems, and [12] for the positive definite Helmholtz case. Our results here also apply to homographic best approximation problems from the steady case, and will thus be useful for the further development of optimized Schwarz methods; we currently study the application to indefinite Helmholtz problems.…”
Abstract. We present and study a homographic best approximation problem, which arises in the analysis of waveform relaxation algorithms with optimized transmission conditions. Its solution characterizes in each class of transmission conditions the one with the best performance of the associated waveform relaxation algorithm. We present the particular class of first order transmission conditions in detail and show that the new waveform relaxation algorithms are well posed and converge much faster than the classical one: the number of iterations to reach a certain accuracy can be orders of magnitudes smaller. We illustrate our analysis with numerical experiments.
“…Similar homographic best approximation problems also occur in the design of optimized Schwarz methods for steady problems, and so far these problems have always been treated by direct analysis; see for example [26,28,27,9] for advection diffusion problems, [7,6,19,16] for indefinite Helmholtz problems, and [12] for the positive definite Helmholtz case. Our results here also apply to homographic best approximation problems from the steady case, and will thus be useful for the further development of optimized Schwarz methods; we currently study the application to indefinite Helmholtz problems.…”
Abstract. We present and study a homographic best approximation problem, which arises in the analysis of waveform relaxation algorithms with optimized transmission conditions. Its solution characterizes in each class of transmission conditions the one with the best performance of the associated waveform relaxation algorithm. We present the particular class of first order transmission conditions in detail and show that the new waveform relaxation algorithms are well posed and converge much faster than the classical one: the number of iterations to reach a certain accuracy can be orders of magnitudes smaller. We illustrate our analysis with numerical experiments.
“…The newest variants in this class of preconditioners is called the sweeping preconditioner [18,19]. Also in domain decomposition there are successful preconditioners, with the first fundamental contribution [16], which then led to optimized Schwarz methods for the Helmholtz equation [29,28]. There are also specialized FETI methods, like FETI-H [25], and FETI-DPH [23], with a convergence analysis in [24].…”
Abstract. We analyze in detail two-grid methods for solving the 1D Helmholtz equation discretized by a standard finite-difference scheme. We explain why both basic components, smoothing and coarse-grid correction, fail for high wave numbers, and show how these components can be modified to obtain a convergent iteration. We show how the parameters of a two-step Jacobi method can be chosen to yield a stable and convergent smoother for the Helmholtz equation. We also stabilize the coarse-grid correction by using a modified wave number determined by dispersion analysis on the coarse grid. Using these modified components we obtain a convergent multigrid iteration for a large range of wave numbers. We also present a complexity analysis which shows that the work scales favorably with the wave number.
“…We also see that it clearly pays to use optimized parameters, as the iteration count is substantially lower than with the first choice of ik in the transmission conditions. We finally show two numerical experiments from [34] and [32], in order to illustrate that optimized Schwarz methods for Helmholtz equations also work well in more practical situations. In Figure 7, we simulated the approach of an Airbus A340 Fig.…”
Section: Domain Decomposition Methods For Helmholtz Problemsmentioning
confidence: 99%
“…In addition, it might be possible to choose an even better transmission condition, as indicated toward the end in Lions' work [51], and also by Hagström et al in [42]. All these observations and further developments led at the turn of the century to the invention of the new class of optimized Schwarz methods [33], with specialized variants for Helmholtz problems [34,32]. For an overview for symmetric coercive problems, see [30].…”
Section: Domain Decomposition Methods For Helmholtz Problemsmentioning
confidence: 99%
“…In addition, to prove convergence for the general overlapping case is an open problem. For the model situation of two subdomains however, one can quantify precisely the dependence of the convergence factor on the wave number k, and the mesh parameter h. We show in Table 5 the resulting convergence factors from [32]. We see that for a fixed wave number k, and a constant overlap, independent of the mesh size h, the algorithm converges with a contraction factor independent of the mesh size h. In the important case where the wave number k scales with the mesh size h like k γ h in order to avoid the pollution effect, see [45,46], we see that the contraction factor only depends very weakly on the growing wave number: for example if the overlap is held constant, all Fourier modes of the error contract at least with a factor 1 − O(k Table 6, we show a numerical experiment for a square cavity open on two sides, and the non-overlapping optimized Schwarz method in order to illustrate the asymptotic results from Table 5.…”
Section: Domain Decomposition Methods For Helmholtz Problemsmentioning
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.