Boundary integral multi-trace formulations and Optimised Schwarz Methods - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles Computers & Mathematics with Applications Year : 2020

Boundary integral multi-trace formulations and Optimised Schwarz Methods


In the present contribution, we consider Helmholtz equation with material coefficients being constants in each subdomain of a geometric partition of the propagation medium (discarding the presence of junctions), and we are interested in the numerical solution of such a problem by means of local multi-trace boundary integral formulations (local-MTF). For a one dimensional problem and configurations with two subdomains, it has been recently established that applying a Jacobi iterative solver to local-MTF is exactly equivalent to an Optimised Schwarz Method (OSM) with a non-local impendance. In the present contribution, we show that this correspondance still holds in the case where the subdomain partition involves an arbitrary number of subdomains. From this, we deduce that the depth of the adjacency graph of the subdomain partition plays a critical role in the convergence of linear solvers applied to local-MTF: we prove it for the case of homogeneous propagation medium and show, through numerical evidences, that this conclusion still holds for heterogeneous media. Our study also shows that, considering variants of local-MTF involving a relaxation parameter, there is a fixed value of this relaxation parameter that systematically leads to optimal speed of convergence for linear solvers.
Fichier principal
Vignette du fichier
manuscript.pdf (1.6 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01921113 , version 1 (13-11-2018)



Xavier Claeys, Pierre Marchand. Boundary integral multi-trace formulations and Optimised Schwarz Methods. Computers & Mathematics with Applications, 2020, 79 (11), ⟨10.1016/j.camwa.2020.01.016⟩. ⟨hal-01921113⟩
128 View
352 Download



Gmail Mastodon Facebook X LinkedIn More