All Title Author
Keywords Abstract


Two-Level Block Decompositions for Solving Helmholtz Equation via Chebyshev Pseudo Spectral Method

DOI: 10.4236/jmp.2018.99107, PP. 1713-1723

Keywords: Helmholtz Equation, Chebyshev Pseudo-Spectral Method, Chebyshev Differentiation Matrix, Coarse-Grain Parallelism, Reflexive Matrix

Full-Text   Cite this paper   Add to My Lib

Abstract:

In this paper, we consider solving the Helmholtz equation \"\" in the Cartesian domain \"\", subject to homogeneous Dirichlet boundary condition, discretized with the Chebyshev pseudo-spectral method. The main purpose of this paper is to present the formulation of a two-level decomposition scheme for decoupling the linear system obtained from the discretization into independent subsystems. This scheme takes advantage of the homogeneity property of the physical problem along one direction to reduce a 2D problem to several 1D problems via a block diagonalization approach and the reflexivity property along the second direction to decompose each of the 1D problems to two independent subproblems using a reflexive decomposition, effectively doubling the number of subproblems. Based on the special structure of the coefficient matrix of the linear system derived from the discretization and a reflexivity property of the second-order Chebyshev differentiation matrix, we show that the decomposed submatrices exhibits a similar property, enabling the system to be decomposed using reflexive decompositions. Explicit forms of the decomposed submatrices are derived. The decomposition not only yields more efficient algorithm but introduces coarse-grain parallelism. Furthermore, it preserves all eigenvalues of the original matrix.

References

[1]  Martinez, J.d.J. and Esperanca, P.d.T.T. (2007) Journal of the Brazilian Society of Mechanical Sciences and Engineering, XXIX, 317-328.
https://doi.org/10.1590/S1678-58782007000300013
[2]  Peyret, R. (2002) Applied Mathematical Sciences. Vol. 148, Springer-Verlag, New York, 448 p.
[3]  Trefethen, L.N. (2000) Spectral Methods in Matlab. SIAM University City Center, Philadelphia, 165 p.
[4]  Chen, H.-C. (2015) Neural, Parallel, and Scientific Computations, 23, 267-275.
[5]  Ehrenstein, U. and Peyret, P. (1989) International Journal for Numerical Methods in Fluids, 9, 427-452.
https://doi.org/10.1002/fld.1650090405
[6]  Golub, G.H. and Van Loan, C.F. (1983) Matrix Computations. The Johns Hopkins University Press, Maryland, 476 p.
[7]  Haidvogel, D.B. and Zang, T. (1979) Journal of Computational Physics, 30, 167-180.
https://doi.org/10.1016/0021-9991(79)90097-4
[8]  Julien, K. and Watson, M. (2009) Journal of Computational Physics, 228, 1480-1503.
https://doi.org/10.1016/j.jcp.2008.10.043
[9]  Lynch, R.E., Rice, J.R. and Thomas, D.H. (1964) Numerische Mathematik, 6, 185-199.
https://doi.org/10.1007/BF01386067
[10]  Chen, H.-C. and Sameh, A. (1989) SIAM Journal on Matrix Analysis and Applications, 10, 39-64.
https://doi.org/10.1137/0610004
[11]  Andrew, A.L. (1973) Technometrics, 15, 405-407.
https://doi.org/10.1080/00401706.1973.10489052
[12]  Cantoni, A. and Butler, P. (1976) Linear Algebra and Its Applications, 13, 275-288.
https://doi.org/10.1016/0024-3795(76)90101-4

Full-Text

comments powered by Disqus

Contact Us

service@oalib.com

QQ:3279437679

微信:OALib Journal