Accelerated simulation methods for plasma kinetics

Russel Caflisch

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Collisional kinetics is a multiscale phenomenon due to the disparity between the continuum (fluid) and the collisional (particle) length scales. This paper describes a class of simulation methods for gases and plasmas, and acceleration techniques for improving their speed and accuracy. Starting from the Landau-Fokker-Planck equation for plasmas, the focus will be on a binary collision model that is solved using a Direct Simulation Monte Carlo (DSMC) method. Acceleration of this method is achieved by coupling the particle method to a continuum fluid description. The velocity distribution function f is represented as a combination of a Maxwellian M (the thermal component) and a set of discrete particles fp (the kinetic component). For systems that are close to (local) equilibrium, this reduces the number N of simulated particles that are required to represent f for a given level of accuracy. We present two methods for exploiting this representation. In the first method, equilibration of particles in fp, as well as disequilibration of particles from M, due to the collision process, is represented by a thermalization/dethermalization step that employs an entropy criterion. Efficiency of the representation is greatly increased by inclusion of particles with negative weights. This significantly complicates the simulation, but the second method is a tractable approach for negatively weighted particles. The accelerated simulation method is compared with standard PIC-DSMC method for both spatially homogeneous problems such as a bump-on-tail and inhomogeneous problems such as nonlinear Landau damping.

Original languageEnglish (US)
Title of host publication30th International Symposium on Rarefied Gas Dynamics, RGD 2016
PublisherAmerican Institute of Physics Inc.
Volume1786
ISBN (Electronic)9780735414488
DOIs
StatePublished - Nov 15 2016
Event30th International Symposium on Rarefied Gas Dynamics, RGD 2016 - Victoria, Canada
Duration: Jul 10 2016Jul 15 2016

Other

Other30th International Symposium on Rarefied Gas Dynamics, RGD 2016
CountryCanada
CityVictoria
Period7/10/167/15/16

Fingerprint

kinetics
simulation
Monte Carlo method
continuums
collisions
Landau damping
fluids
Fokker-Planck equation
velocity distribution
distribution functions
inclusions
entropy
gases

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Caflisch, R. (2016). Accelerated simulation methods for plasma kinetics. In 30th International Symposium on Rarefied Gas Dynamics, RGD 2016 (Vol. 1786). [020001] American Institute of Physics Inc.. https://doi.org/10.1063/1.4967537

Accelerated simulation methods for plasma kinetics. / Caflisch, Russel.

30th International Symposium on Rarefied Gas Dynamics, RGD 2016. Vol. 1786 American Institute of Physics Inc., 2016. 020001.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Caflisch, R 2016, Accelerated simulation methods for plasma kinetics. in 30th International Symposium on Rarefied Gas Dynamics, RGD 2016. vol. 1786, 020001, American Institute of Physics Inc., 30th International Symposium on Rarefied Gas Dynamics, RGD 2016, Victoria, Canada, 7/10/16. https://doi.org/10.1063/1.4967537
Caflisch R. Accelerated simulation methods for plasma kinetics. In 30th International Symposium on Rarefied Gas Dynamics, RGD 2016. Vol. 1786. American Institute of Physics Inc. 2016. 020001 https://doi.org/10.1063/1.4967537
Caflisch, Russel. / Accelerated simulation methods for plasma kinetics. 30th International Symposium on Rarefied Gas Dynamics, RGD 2016. Vol. 1786 American Institute of Physics Inc., 2016.
@inproceedings{b5d44882c6d14e77aed320ccdc605248,
title = "Accelerated simulation methods for plasma kinetics",
abstract = "Collisional kinetics is a multiscale phenomenon due to the disparity between the continuum (fluid) and the collisional (particle) length scales. This paper describes a class of simulation methods for gases and plasmas, and acceleration techniques for improving their speed and accuracy. Starting from the Landau-Fokker-Planck equation for plasmas, the focus will be on a binary collision model that is solved using a Direct Simulation Monte Carlo (DSMC) method. Acceleration of this method is achieved by coupling the particle method to a continuum fluid description. The velocity distribution function f is represented as a combination of a Maxwellian M (the thermal component) and a set of discrete particles fp (the kinetic component). For systems that are close to (local) equilibrium, this reduces the number N of simulated particles that are required to represent f for a given level of accuracy. We present two methods for exploiting this representation. In the first method, equilibration of particles in fp, as well as disequilibration of particles from M, due to the collision process, is represented by a thermalization/dethermalization step that employs an entropy criterion. Efficiency of the representation is greatly increased by inclusion of particles with negative weights. This significantly complicates the simulation, but the second method is a tractable approach for negatively weighted particles. The accelerated simulation method is compared with standard PIC-DSMC method for both spatially homogeneous problems such as a bump-on-tail and inhomogeneous problems such as nonlinear Landau damping.",
author = "Russel Caflisch",
year = "2016",
month = "11",
day = "15",
doi = "10.1063/1.4967537",
language = "English (US)",
volume = "1786",
booktitle = "30th International Symposium on Rarefied Gas Dynamics, RGD 2016",
publisher = "American Institute of Physics Inc.",

}

TY - GEN

T1 - Accelerated simulation methods for plasma kinetics

AU - Caflisch, Russel

PY - 2016/11/15

Y1 - 2016/11/15

N2 - Collisional kinetics is a multiscale phenomenon due to the disparity between the continuum (fluid) and the collisional (particle) length scales. This paper describes a class of simulation methods for gases and plasmas, and acceleration techniques for improving their speed and accuracy. Starting from the Landau-Fokker-Planck equation for plasmas, the focus will be on a binary collision model that is solved using a Direct Simulation Monte Carlo (DSMC) method. Acceleration of this method is achieved by coupling the particle method to a continuum fluid description. The velocity distribution function f is represented as a combination of a Maxwellian M (the thermal component) and a set of discrete particles fp (the kinetic component). For systems that are close to (local) equilibrium, this reduces the number N of simulated particles that are required to represent f for a given level of accuracy. We present two methods for exploiting this representation. In the first method, equilibration of particles in fp, as well as disequilibration of particles from M, due to the collision process, is represented by a thermalization/dethermalization step that employs an entropy criterion. Efficiency of the representation is greatly increased by inclusion of particles with negative weights. This significantly complicates the simulation, but the second method is a tractable approach for negatively weighted particles. The accelerated simulation method is compared with standard PIC-DSMC method for both spatially homogeneous problems such as a bump-on-tail and inhomogeneous problems such as nonlinear Landau damping.

AB - Collisional kinetics is a multiscale phenomenon due to the disparity between the continuum (fluid) and the collisional (particle) length scales. This paper describes a class of simulation methods for gases and plasmas, and acceleration techniques for improving their speed and accuracy. Starting from the Landau-Fokker-Planck equation for plasmas, the focus will be on a binary collision model that is solved using a Direct Simulation Monte Carlo (DSMC) method. Acceleration of this method is achieved by coupling the particle method to a continuum fluid description. The velocity distribution function f is represented as a combination of a Maxwellian M (the thermal component) and a set of discrete particles fp (the kinetic component). For systems that are close to (local) equilibrium, this reduces the number N of simulated particles that are required to represent f for a given level of accuracy. We present two methods for exploiting this representation. In the first method, equilibration of particles in fp, as well as disequilibration of particles from M, due to the collision process, is represented by a thermalization/dethermalization step that employs an entropy criterion. Efficiency of the representation is greatly increased by inclusion of particles with negative weights. This significantly complicates the simulation, but the second method is a tractable approach for negatively weighted particles. The accelerated simulation method is compared with standard PIC-DSMC method for both spatially homogeneous problems such as a bump-on-tail and inhomogeneous problems such as nonlinear Landau damping.

UR - http://www.scopus.com/inward/record.url?scp=85008873286&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85008873286&partnerID=8YFLogxK

U2 - 10.1063/1.4967537

DO - 10.1063/1.4967537

M3 - Conference contribution

AN - SCOPUS:85008873286

VL - 1786

BT - 30th International Symposium on Rarefied Gas Dynamics, RGD 2016

PB - American Institute of Physics Inc.

ER -