|
|
Sparse Jacobian Construction for Mapped Grid Visco-Resistive Magnetohydrodynamics-
incollection
- | |
|
Author(s)
Daniel R. Reynolds
, Ravi Samtaney
|
Published in Recent Advances in Algorithmic Differentiation
|
Editor(s) Shaun Forth, Paul Hovland, Eric Phipps, Jean Utke, Andrea Walther |
Year 2012 |
Publisher Springer |
Abstract We apply the automatic differentiation tool OpenAD toward constructing a preconditioner for fully implicit simulations of mapped grid visco-resistive magnetohydrodynamics (MHD), used in modeling tokamak fusion devices. Our simulation framework employs a fully implicit formulation in time, and a mapped finite volume spatial discretization. We solve this model using inexact Newton-Krylov methods. Of critical importance in these iterative solvers is the development of an effective preconditioner, which typically requires knowledge of the Jacobian of the nonlinear residual function. However, due to significant nonlinearity within our PDE system, our mapped spatial discretization, and stencil adaptivity at physical boundaries, analytical derivation of these Jacobian entries is highly nontrivial. This paper therefore focuses on Jacobian construction using automatic differentiation. In particular, we discuss applying OpenAD to the case of a spatially-adaptive stencil patch that automatically handles differences between the domain interior and boundary, and configuring ad for reduced stencil approximations to the Jacobian. We investigate both scalar and vector tangent mode differentiation, along with simple finite difference approaches, to compare the resulting accuracy and efficiency of Jacobian construction in this application. |
Cross-References Forth2012RAi |
AD Tools OpenAD |
AD Theory and Techniques Sparsity |
BibTeX
@INCOLLECTION{
Reynolds2012SJC,
title = "Sparse {J}acobian Construction for Mapped Grid Visco-Resistive Magnetohydrodynamics",
doi = "10.1007/978-3-642-30023-3_2",
author = "Daniel R. Reynolds and Ravi Samtaney",
abstract = "We apply the automatic differentiation tool OpenAD toward constructing a
preconditioner for fully implicit simulations of mapped grid visco-resistive magnetohydrodynamics
(MHD), used in modeling tokamak fusion devices. Our simulation framework employs a fully implicit
formulation in time, and a mapped finite volume spatial discretization. We solve this model using
inexact Newton-Krylov methods. Of critical importance in these iterative solvers is the development
of an effective preconditioner, which typically requires knowledge of the Jacobian of the nonlinear
residual function. However, due to significant nonlinearity within our PDE system, our mapped
spatial discretization, and stencil adaptivity at physical boundaries, analytical derivation of
these Jacobian entries is highly nontrivial. This paper therefore focuses on Jacobian construction
using automatic differentiation. In particular, we discuss applying OpenAD to the case of a
spatially-adaptive stencil patch that automatically handles differences between the domain interior
and boundary, and configuring AD for reduced stencil approximations to the Jacobian. We investigate
both scalar and vector tangent mode differentiation, along with simple finite difference approaches,
to compare the resulting accuracy and efficiency of Jacobian construction in this application.",
pages = "11--21",
crossref = "Forth2012RAi",
booktitle = "Recent Advances in Algorithmic Differentiation",
series = "Lecture Notes in Computational Science and Engineering",
publisher = "Springer",
address = "Berlin",
volume = "87",
editor = "Shaun Forth and Paul Hovland and Eric Phipps and Jean Utke and Andrea Walther",
isbn = "978-3-540-68935-5",
issn = "1439-7358",
year = "2012",
ad_tools = "OpenAD",
ad_theotech = "Sparsity"
}
| |
back
|
|