In this paper we report the convergence acceleration effect of the extended node patch preconditioner for the iterative full-wave electromagnetic finite element method with more than ten million degrees of freedom. The preconditioner, which is categorized into the multiplicative Schwarz scheme, effectively works with conventional numerical iterative matrix solving methods on a parallel computer. We examined the convergence properties of the preconditioner combined with the COCG, COCR and GMRES algorithms for the analysis domain encompassed by absorbing boundary conditions (ABC) such as perfectly matched layers (PML). In those analyses the properties of the convergence are investigated numerically by sweeping frequency range and the number of PMLs. Memory-efficient nature of the preconditioner is numerically confirmed through the experiments and upper bounds of the required memory size are theoretically proved. Finally it is demonstrated that this extended node patch preconditioner with GMRES algorithm works well with the problems up to one hundred million degrees of freedom.
Toshio MURAYAMA
Sony Global Manufacturing & Operations Corporation
Akira MUTO
Sony Global Manufacturing & Operations Corporation
Amane TAKEI
University of Miyazaki
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Toshio MURAYAMA, Akira MUTO, Amane TAKEI, "Convergence Properties of Iterative Full-Wave Electromagnetic FEM Analyses with Node Block Preconditioners" in IEICE TRANSACTIONS on Electronics,
vol. E101-C, no. 8, pp. 612-619, August 2018, doi: 10.1587/transele.E101.C.612.
Abstract: In this paper we report the convergence acceleration effect of the extended node patch preconditioner for the iterative full-wave electromagnetic finite element method with more than ten million degrees of freedom. The preconditioner, which is categorized into the multiplicative Schwarz scheme, effectively works with conventional numerical iterative matrix solving methods on a parallel computer. We examined the convergence properties of the preconditioner combined with the COCG, COCR and GMRES algorithms for the analysis domain encompassed by absorbing boundary conditions (ABC) such as perfectly matched layers (PML). In those analyses the properties of the convergence are investigated numerically by sweeping frequency range and the number of PMLs. Memory-efficient nature of the preconditioner is numerically confirmed through the experiments and upper bounds of the required memory size are theoretically proved. Finally it is demonstrated that this extended node patch preconditioner with GMRES algorithm works well with the problems up to one hundred million degrees of freedom.
URL: https://global.ieice.org/en_transactions/electronics/10.1587/transele.E101.C.612/_p
Copy
@ARTICLE{e101-c_8_612,
author={Toshio MURAYAMA, Akira MUTO, Amane TAKEI, },
journal={IEICE TRANSACTIONS on Electronics},
title={Convergence Properties of Iterative Full-Wave Electromagnetic FEM Analyses with Node Block Preconditioners},
year={2018},
volume={E101-C},
number={8},
pages={612-619},
abstract={In this paper we report the convergence acceleration effect of the extended node patch preconditioner for the iterative full-wave electromagnetic finite element method with more than ten million degrees of freedom. The preconditioner, which is categorized into the multiplicative Schwarz scheme, effectively works with conventional numerical iterative matrix solving methods on a parallel computer. We examined the convergence properties of the preconditioner combined with the COCG, COCR and GMRES algorithms for the analysis domain encompassed by absorbing boundary conditions (ABC) such as perfectly matched layers (PML). In those analyses the properties of the convergence are investigated numerically by sweeping frequency range and the number of PMLs. Memory-efficient nature of the preconditioner is numerically confirmed through the experiments and upper bounds of the required memory size are theoretically proved. Finally it is demonstrated that this extended node patch preconditioner with GMRES algorithm works well with the problems up to one hundred million degrees of freedom.},
keywords={},
doi={10.1587/transele.E101.C.612},
ISSN={1745-1353},
month={August},}
Copy
TY - JOUR
TI - Convergence Properties of Iterative Full-Wave Electromagnetic FEM Analyses with Node Block Preconditioners
T2 - IEICE TRANSACTIONS on Electronics
SP - 612
EP - 619
AU - Toshio MURAYAMA
AU - Akira MUTO
AU - Amane TAKEI
PY - 2018
DO - 10.1587/transele.E101.C.612
JO - IEICE TRANSACTIONS on Electronics
SN - 1745-1353
VL - E101-C
IS - 8
JA - IEICE TRANSACTIONS on Electronics
Y1 - August 2018
AB - In this paper we report the convergence acceleration effect of the extended node patch preconditioner for the iterative full-wave electromagnetic finite element method with more than ten million degrees of freedom. The preconditioner, which is categorized into the multiplicative Schwarz scheme, effectively works with conventional numerical iterative matrix solving methods on a parallel computer. We examined the convergence properties of the preconditioner combined with the COCG, COCR and GMRES algorithms for the analysis domain encompassed by absorbing boundary conditions (ABC) such as perfectly matched layers (PML). In those analyses the properties of the convergence are investigated numerically by sweeping frequency range and the number of PMLs. Memory-efficient nature of the preconditioner is numerically confirmed through the experiments and upper bounds of the required memory size are theoretically proved. Finally it is demonstrated that this extended node patch preconditioner with GMRES algorithm works well with the problems up to one hundred million degrees of freedom.
ER -