TY - JOUR
T1 - High order numerical simulation of the transmission and scattering of waves using the method of difference potentials
AU - Medvinsky, M.
AU - Tsynkov, S.
AU - Turkel, E.
PY - 2013/6/5
Y1 - 2013/6/5
N2 - The method of difference potentials generalizes the method of Calderon's operators from PDEs to arbitrary difference equations and systems. It offers several key advantages, such as the capability of handling boundaries/interfaces that are not aligned with the discretization grid, variable coefficients, and nonstandard boundary conditions. In doing so, the complexity of the algorithm remains comparable to that of an ordinary finite difference scheme on a regular structured grid.Previously, we have applied the method of difference potentials to solving several variable coefficient interior Helmholtz problems with fourth and sixth order accuracy. We have employed compact finite difference schemes as a core discretization methodology. Those schemes enable high order accuracy on narrow stencils and hence require only as many boundary conditions as needed for the underlying differential equation itself. Numerical experiments corroborate the high order accuracy of our method for variable coefficients, regular grids, and non-conforming boundaries.In the current paper, we extend the previously developed methodology to exterior problems. We present a complete theoretical analysis of the algorithm, as well as the results of a series of numerical simulations. Specifically, we study the scattering of time-harmonic waves about smooth shapes, subject to various boundary conditions. We also solve the transmission/scattering problems, in which not only do the waves scatter off a given shape but also propagate through the interface and travel across the heterogeneous medium inside. In all the cases, our methodology guarantees high order accuracy for regular grids and non-conforming boundaries and interfaces.
AB - The method of difference potentials generalizes the method of Calderon's operators from PDEs to arbitrary difference equations and systems. It offers several key advantages, such as the capability of handling boundaries/interfaces that are not aligned with the discretization grid, variable coefficients, and nonstandard boundary conditions. In doing so, the complexity of the algorithm remains comparable to that of an ordinary finite difference scheme on a regular structured grid.Previously, we have applied the method of difference potentials to solving several variable coefficient interior Helmholtz problems with fourth and sixth order accuracy. We have employed compact finite difference schemes as a core discretization methodology. Those schemes enable high order accuracy on narrow stencils and hence require only as many boundary conditions as needed for the underlying differential equation itself. Numerical experiments corroborate the high order accuracy of our method for variable coefficients, regular grids, and non-conforming boundaries.In the current paper, we extend the previously developed methodology to exterior problems. We present a complete theoretical analysis of the algorithm, as well as the results of a series of numerical simulations. Specifically, we study the scattering of time-harmonic waves about smooth shapes, subject to various boundary conditions. We also solve the transmission/scattering problems, in which not only do the waves scatter off a given shape but also propagate through the interface and travel across the heterogeneous medium inside. In all the cases, our methodology guarantees high order accuracy for regular grids and non-conforming boundaries and interfaces.
KW - Artificial boundary conditions
KW - Boundary projections
KW - Calderon's operators
KW - Compact differencing
KW - Curvilinear boundaries
KW - Difference potentials
KW - Exterior problems
KW - High order accuracy
KW - Non-conforming boundaries/interfaces
KW - Regular grids
KW - Variable coefficients
UR - http://www.scopus.com/inward/record.url?scp=84876302200&partnerID=8YFLogxK
U2 - 10.1016/j.jcp.2013.03.014
DO - 10.1016/j.jcp.2013.03.014
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:84876302200
SN - 0021-9991
VL - 243
SP - 305
EP - 322
JO - Journal of Computational Physics
JF - Journal of Computational Physics
ER -