Difference between revisions of "Domain decomposition methods for waves"

From ONELAB
Jump to: navigation, search
Line 3: Line 3:
 
== Introduction ==
 
== Introduction ==
  
To run the model, open '''waveguide3d.pro''' with Gmsh.
+
To run the model, open '''waveguide3d.pro''' with Gmsh. (This is for demonstration uses only. For actual, parallel computations, you should [https://geuz.org/trac/getdp/wiki/ForDavid compile GetDP and Gmsh with MPI support] and [http://onelab.info/files/ddm_wave_simple/run_slurm.sh run the examples from the command line on an HPC cluster].
  
 
The formulations implement non-overlapping Schwarz domain decomposition methods for the Helmholtz equation and for the time-harmonic Maxwell system. Several families of transmission conditions are implemented: zeroth- and second-order optimized conditions, as well as new Padé-localized square-root conditions<ref name=BoubAntGeuz2012 />.
 
The formulations implement non-overlapping Schwarz domain decomposition methods for the Helmholtz equation and for the time-harmonic Maxwell system. Several families of transmission conditions are implemented: zeroth- and second-order optimized conditions, as well as new Padé-localized square-root conditions<ref name=BoubAntGeuz2012 />.
 +
 +
  
 
== References ==
 
== References ==

Revision as of 20:19, 21 May 2014

Optimized Schwarz domain decomposition methods for time-harmonic wave problems

Download model archive (ddm_wave_simple.zip)
Browse individual model files

Introduction

To run the model, open waveguide3d.pro with Gmsh. (This is for demonstration uses only. For actual, parallel computations, you should compile GetDP and Gmsh with MPI support and run the examples from the command line on an HPC cluster.

The formulations implement non-overlapping Schwarz domain decomposition methods for the Helmholtz equation and for the time-harmonic Maxwell system. Several families of transmission conditions are implemented: zeroth- and second-order optimized conditions, as well as new Padé-localized square-root conditions[1].


References

  1. Y. Boubendir, X. Antoine, C. Geuzaine. A quasi-optimal non-overlapping domain decomposition algorithm for the Helmholtz equation. Journal of Computational Physics 231 (2), 262-280, 2012.


Models developed by X. Antoine, Y. Boubendir, M. El Bouajaji, D. Colignon, C. Geuzaine, N. Marsic, B. Thierry, S. Tournier and A. Vion. This work was funded in part by the Belgian Science Policy (IAP P6/21 and P7/02), the Belgian French Community (ARC 09/14-02), the Walloon Region (WIST3 No 1017086 ONELAB and ALIZEES), the Agence Nationale pour la Recherche (ANR-09-BLAN-0057-01 MicroWave) and the EADS Foundation (grant 089-1009-1006 High-BRID).