Load balancing in electromagnetic field computation using distributed systems

• Autorzy: Szymański, G. , Waszak, M.
• Języki publikacji: EN

Abstrakty

EN

This paper deals with electromagnetic field computations using distributed systems. Field problems given by Maxwell's partial differential equations are solved using well-established methods under given initial and boundary conditions. Spatial discretization performed by using the finite element method leads to linear systems of large dimension. Due to the large solution time, linear algebra methods are applied in parallel. Parallelism of conjugate gradient algorithm is considered and decomposition of computational domain amongst processors is discussed. It is focused on both load balancing and communication effort minimisation. An adaptive approach, based on recursive bisection, for load balancing is proposed and results obtained on a cluster of workstations are presented.

Słowa kluczowe

PL

system rozproszony; pole elektromagnetyczne

EN

load balancing; electromagnetic field; distributed system

• Czasopismo: Foundations of Computing and Decision Sciences
• Rocznik: 2003
• Tom: Vol. 28, No. 3
• Strony: 179-189
• Opis fizyczny: Bibliogr. 15 poz.

Twórcy

autor

Szymański, G.

• Institute of Control and System Engineering, Poznan University of Technology, ul. Piotrowo 3A, 60-965 Poznan, Poland

autor

Waszak, M.

• Institute of Control and System Engineering, Poznan University of Technology, ul. Piotrowo 3A, 60-965 Poznan, Poland

Bibliografia

• [1] Barrett R. et al., Templates for the Solution of Linear Systems: Building Blocks for Iterative Methods, SIAM, Philadelphia, PA, 1994.
• [2] Cagniot E., Brandes T., Dekeyser J. L., Piriou F., Parallelization of a 3D Magnetostatic Code Using High Performance Fortran with the Schur Complement Method, Conference Record of the 13^th COMPUMAG Conference on the Computation of Electromagnetic Fields, Evian, 2001, 10-11.
• [3] Chari M., Bedrosian G., D'Angelo J., Konrad A., Finite element applications in electrical engineering, IEEE Transactions on Magnetics, 29, 2, 1993, 1306-1313.
• [4] Demmel J., Heath M., Van der Vorst H., Parallel Numerical Linear Algebra, Acta Numerica, 2, 1993, 111-197.
• [5] Foster I. T., Designing and Building Parallel Programs, Addison-Wesley, 1995.
• [6] Fox G., Wiliams R., Messina P., Parallel Computing Works!, Morgan Kaufmann Publishers, San Francisco, 1994.
• [7] Geist A. et al., Parallel Virtual Machine - A User's Guide and Tutorial for Networked Parallel Computing, MIT Press, 1994.
• [8] Hunter P., Pullan A., FEM/BEM Notes, The University of Auckland, 2001.
• [9] Im E.-J., Optimizing of the Performance of Sparse Matrix-Vector Multiplication, PhD thesis, University of California at Berkeley, 2000.
• [10] Iwashita T., Shimasaki M., Construction and Ordering of Edge Elements for Parallel Computation, IEEE Transactions on Magnetics, 37, 5, 2001, 3498-3502.
• [11] Patecki A., Simulation of quasi steady-state electrodynamic phenomena using finite difference method, Dissertation Nr. 345, Poznan University of Technology Press, Poznan, 1999.
• [12] Patecki A., Szymanski G., Calculation of 3D eddy current problems by the finite difference method using the A-V formulation, Conference Record of the Six Biennal IEEE Conference on Electromagnetic Field Computation, Aix-les-Bains, 1994, 18.
• [13] Saad Y., Iterative Methods for Sparse Linear Systems, PWS Publishing Company, Boston, 2000.
• [14] Taflove A., Computational Electrodynamics: The Finite-Difference Time-Domain Method, Artech House, 1995.
• [15] Van der Vorst H., Dekker K., Conjugate gradient type methods and preconditioning, Journal of Computational and Applied Mathematics, 24, 1988, 73-87.