Investigacíon Operativa, Vol.4, No.1, pp.27-48, 1994.
I.F. M. Menezes,
Department of Civil Engineering, PUC-Rio, 22453, Rio de Janeiro, Brazil.
G.H. Paulino,
School of Civil and Environmental Engineering, Cornell University, Ithaca, NY 14853,
U.S.A.
M. Gattass,
Department of Computer Science, PUC-Rio, 22453, Rio de Janeiro, Brazil.
S. Mukherjee
Department of Theoretical and Applied Mechanics, Cornell University, Ithaca, NY 14853,
U.S.A.
Abstract
This paper investigates profile and wavefront reduction of large sparse matrices of symmetric configuration. A new version of the Spectral Finite Element Graph (FEG) Resequencing (SFR) algorithm is presented. This version, based on the Sturm Sequence Method (SFRSturm), is compared with previous implementation based on the Subspace Iteration Method (SIM), and is denoted here as SFRSIM. SFR type algorithms are effective to reorder vertices of generic graphs. These vertices can be associated to nodes or elements of finite element meshes. Several numerical examples are presented here. The numerical experiments include evaluation of algorithm performance using practical finite element meshes, a brief comparison between the SFRSturm and SFRSIM implementations, and assessment of the SFRSturm algorithm with respect to factorization of the associated system matrix. Moreover, the SFRSturm is also compared with previous resequencing algorithms, which are available in the technical literature.
Key words: sparse matrix, resequencing algorithms, profile and wavefront reduction, eigenpair of graphs, Laplacian matrix, Sturm sequence, subspace iteration.
Representative Results:
|
|
Offshore Structure and Turbine Blade |