A proposal for pseudospectra computation on graphic processor units
Article Sidebar
Main Article Content
Abstract
Introduction. Computation of matrix pseudospectra is required in many applications when modeling by differential equations. This computation is really expensive, especially for large matrices, for which highly parallelizable algorithms have been successfully implemented on high performance computers. Objective. We present an exploratory analysis of the pseudospectra computation in a hybrid architecture CPU-GPU where the graphics processing unit performs the massive parallel computation. Methodology. A proposal is formulated after analyzing some parallel implementations on high performance computers, methods based on Krylov methods, and the capacities of the graphics processor units for massive computation in the large-scale setting. Results. The proposal is attractive since the graphics processing units currently can be found on a wide range of computers, or can be adapted to any computer at a very a low cost. Conclusions. In this document we describe a general scheme for the parallel computation of pseudospectra on a hybrid architecture CPU-GPU.
Downloads
Article Details
References
Bekas, C., Kokiopoulou, E., Koutis, I., Gallopoulus, E. (2001) Towards the effective Parallel Computation of Matrix Pseudospectra. IC’s 01: Proceeding of the 15th International Conference on Supercomputing, June 2001, 203-226.
Bekas, C., Kokiopoulou, E., Gallopoulos E., Simoncini, V., Parallel Computation of Pseudospectra using Transfer Functions on a MATLAB-MPI Cluster Platform. Lecture Notes in Computer Science. DOI:10.1007/3-540-45825-5_35.
Bell, N., Garland, M. (2009). Implementing sparse matrix-vector multiplication on throughput-oriented processors. SC’09: Proceeding of the Conference on High Performance Computing Networking. Storage and Analysis, New York, NY, USA (ACM), 1-11.
Bell, N., Hoberock, J. (2011). Thrust: A Productivity-Oriented Library for CUDA. GPU Computing Gems, Jade Edition. Editores Wen-mei W. Hwu.
Castillo, Z. (2004). A new algorithm for continuation and bifurcation analysis of large-scale free surface flows. PhD thesis, Rice University, Houston, Texas.
Embree, M., Trefethen. L. N., Pseudospectra Gateway (2021). [Online]. Disponible en: https://www.cs.ox.ac.uk20/pseudospectra/index.html.
Guevara, R. (2012). Paralelización de datos para el cálculo del pseudoespectro. Tesis de Licenciatura. Universidad Central de Venezuela, Escuela de Computación, Caracas, Venezuela.
Golub, G., & Van Loan, C. (1996). Matrix Computations (3rd ed.). The Johns Hopkins University.
Lui, S. (1997). Computation of pseudospectra by continuation. SIAM J. Sci. Comput., 18, 2(1997), 565–573.
Mezher, D., Philippe, B. (2002). Parallel computation of pseudospectra of large sparse matrices. Parallel Computing, 28, 199-221.
Minini, P., Rosenberg, D., Reddy, R., Pouquet, A. (2011). A hybrid MPI-OpenMP scheme for scalable parallel pseudospectral computations for fluid turbulence. Parallel Computing, 37(6-7), 316-312.
Noschese, S. & Reichel, L. (2015). Approximated structured pseudospectra. Numerical Linear Algebra and Applications, 00, 1-15.
NVIDIA Corporation, NVIDIA CUDA C Programming Guide (Version 11.3.0) (2021). [Online]. Disponible en: https://docs.nvidia.com/cuda/cuda-c-programming-guide
Otero, B., Astudillo, R., Castillo, Z. (2015). Un esquema paralelo para el cálculo del pseudoespectro de matrices de gran magnitud. Revista Internacional de Métodos Numéricos para el Cálculo y Diseño en Ingeniería. 31-1, 8-12.
Sorensen D. C. (1997). Implicitly restarted Arnoldi/Lanczos methods for large scale eigenvalue calculations, Parallel Numerical Algorithms, Springer, Dordrecht, 119-165.
Trefethen, L. N. & Bau III, D. (1997). Numerical linear algebra. Siam, Philadelphia.
Trefethen, L.N. & Embree, M. (2005). Spectra and Pseudospectra: the behavior of nonnormal Matrices and Operators, Princeton University Press, New Jersey, USA.
Trefethen L.N & Wright, T. (2001). Large-scale computation of pseudospectra using ARPACK and eigs, SIAM J. Sci. Comput. 23 (2001), 591–605.
Trefethen, L.N. (1999). Computation of pseudospectra. Acta Numerica 8, 1, 247–295.