Parallelization of a Vlasov–Maxwell solver in four-dimensional phase space

Lars K. S. Daldorff, Bengt Eliasson

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

We present a parallelized algorithm for solving the time-dependent Vlasov–Maxwell system of equations in the four-dimensional phase space (two spatial and velocity dimensions). One Vlasov equation is solved for each particle species, from which charge and current densities are calculated for the Maxwell equations. The parallelization is divided into two different layers. For the first layer, each plasma species is given its own processor group. On the second layer, the distribution function is domain decomposed on its dedicated resources. By separating the communication and calculation steps, we have met the design criteria of good speedup and simplicity in the implementation.
LanguageEnglish
Pages109-115
Number of pages7
JournalParallel Computing
Volume35
Issue number2
DOIs
Publication statusPublished - Feb 2009

Fingerprint

Vlasov equation
Maxwell equations
Charge density
Parallelization
Distribution functions
Phase Space
Current density
Plasmas
plasma layers
vlasov equations
Communication
Maxwell equation
central processing units
Vlasov Equation
Maxwell System
resources
communication
distribution functions
current density
Maxwell's equations

Keywords

  • numerical simulations
  • vlsaov equation
  • parallel algorithm

Cite this

@article{02ed5e50d202448cba78722662415b2b,
title = "Parallelization of a Vlasov–Maxwell solver in four-dimensional phase space",
abstract = "We present a parallelized algorithm for solving the time-dependent Vlasov–Maxwell system of equations in the four-dimensional phase space (two spatial and velocity dimensions). One Vlasov equation is solved for each particle species, from which charge and current densities are calculated for the Maxwell equations. The parallelization is divided into two different layers. For the first layer, each plasma species is given its own processor group. On the second layer, the distribution function is domain decomposed on its dedicated resources. By separating the communication and calculation steps, we have met the design criteria of good speedup and simplicity in the implementation.",
keywords = "numerical simulations, vlsaov equation, parallel algorithm",
author = "Daldorff, {Lars K. S.} and Bengt Eliasson",
year = "2009",
month = "2",
doi = "10.1016/j.parco.2008.11.002",
language = "English",
volume = "35",
pages = "109--115",
journal = "Parallel Computing",
issn = "0167-8191",
number = "2",

}

Parallelization of a Vlasov–Maxwell solver in four-dimensional phase space. / Daldorff, Lars K. S.; Eliasson, Bengt.

In: Parallel Computing, Vol. 35, No. 2, 02.2009, p. 109-115.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Parallelization of a Vlasov–Maxwell solver in four-dimensional phase space

AU - Daldorff, Lars K. S.

AU - Eliasson, Bengt

PY - 2009/2

Y1 - 2009/2

N2 - We present a parallelized algorithm for solving the time-dependent Vlasov–Maxwell system of equations in the four-dimensional phase space (two spatial and velocity dimensions). One Vlasov equation is solved for each particle species, from which charge and current densities are calculated for the Maxwell equations. The parallelization is divided into two different layers. For the first layer, each plasma species is given its own processor group. On the second layer, the distribution function is domain decomposed on its dedicated resources. By separating the communication and calculation steps, we have met the design criteria of good speedup and simplicity in the implementation.

AB - We present a parallelized algorithm for solving the time-dependent Vlasov–Maxwell system of equations in the four-dimensional phase space (two spatial and velocity dimensions). One Vlasov equation is solved for each particle species, from which charge and current densities are calculated for the Maxwell equations. The parallelization is divided into two different layers. For the first layer, each plasma species is given its own processor group. On the second layer, the distribution function is domain decomposed on its dedicated resources. By separating the communication and calculation steps, we have met the design criteria of good speedup and simplicity in the implementation.

KW - numerical simulations

KW - vlsaov equation

KW - parallel algorithm

UR - http://www.sciencedirect.com/science/article/pii/S0167819108001336

U2 - 10.1016/j.parco.2008.11.002

DO - 10.1016/j.parco.2008.11.002

M3 - Article

VL - 35

SP - 109

EP - 115

JO - Parallel Computing

T2 - Parallel Computing

JF - Parallel Computing

SN - 0167-8191

IS - 2

ER -