Advanced

Adopting (s)EPIRK schemes in a domain-based IMEX setting

Straub, Veronika; Ortleb, Sigrun; Birken, Philipp LU and Meister, Andreas (2017) International Conference of Numerical Analysis and Applied Mathematics 2016, ICNAAM 2016 In Proceedings of ICNAAM 2016, AIP Conference Proceedings 1863 410008.
Abstract

The simulation of viscous, compressible flows around complex geometries or similar applications often inherit the task of solving large, stiff systems of ODEs. Domain-based implicit-explicit (IMEX) type schemes offer the possibility to apply two different schemes to different parts of the computational domain. The goal hereby is to decrease the computational cost by increasing the admissible step sizes with no loss of stability and by reducing the system sizes of the linear solver within the implicit integrator. But which combination of methods reaches the largest gain in efficiency? Coupling of Runge-Kutta methods or different multistep methods has been investigated so far by other authors. Here, we inspect the adoption of the recently... (More)

The simulation of viscous, compressible flows around complex geometries or similar applications often inherit the task of solving large, stiff systems of ODEs. Domain-based implicit-explicit (IMEX) type schemes offer the possibility to apply two different schemes to different parts of the computational domain. The goal hereby is to decrease the computational cost by increasing the admissible step sizes with no loss of stability and by reducing the system sizes of the linear solver within the implicit integrator. But which combination of methods reaches the largest gain in efficiency? Coupling of Runge-Kutta methods or different multistep methods has been investigated so far by other authors. Here, we inspect the adoption of the recently introduced exponential integrators called EPIRK and sEPIRK in the IMEX setting, since they are perfectly suited for large, stiff systems of ODEs.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
in
Proceedings of ICNAAM 2016, AIP Conference Proceedings 1863
volume
410008
publisher
American Institute of Physics Inc.
conference name
International Conference of Numerical Analysis and Applied Mathematics 2016, ICNAAM 2016
external identifiers
  • scopus:85026635946
  • scopus:85026635946
  • wos:000410159800409
ISBN
9780735415386
DOI
10.1063/1.4992588
language
English
LU publication?
yes
id
cc8dcfd6-28e0-4acf-ba6d-98952bc89995
date added to LUP
2017-07-26 14:32:48
date last changed
2018-01-16 13:21:51
@inproceedings{cc8dcfd6-28e0-4acf-ba6d-98952bc89995,
  abstract     = {<p>The simulation of viscous, compressible flows around complex geometries or similar applications often inherit the task of solving large, stiff systems of ODEs. Domain-based implicit-explicit (IMEX) type schemes offer the possibility to apply two different schemes to different parts of the computational domain. The goal hereby is to decrease the computational cost by increasing the admissible step sizes with no loss of stability and by reducing the system sizes of the linear solver within the implicit integrator. But which combination of methods reaches the largest gain in efficiency? Coupling of Runge-Kutta methods or different multistep methods has been investigated so far by other authors. Here, we inspect the adoption of the recently introduced exponential integrators called EPIRK and sEPIRK in the IMEX setting, since they are perfectly suited for large, stiff systems of ODEs.</p>},
  author       = {Straub, Veronika and Ortleb, Sigrun and Birken, Philipp and Meister, Andreas},
  booktitle    = {Proceedings of ICNAAM 2016, AIP Conference Proceedings 1863},
  isbn         = {9780735415386},
  language     = {eng},
  month        = {07},
  publisher    = {American Institute of Physics Inc.},
  title        = {Adopting (s)EPIRK schemes in a domain-based IMEX setting},
  url          = {http://dx.doi.org/10.1063/1.4992588},
  volume       = {410008},
  year         = {2017},
}