Parallel algorithms for the solution of a sparse system

✍ Scribed by Jenn-Ching Luo

• Publisher: Elsevier Science
• Year: 1993
• Weight: 612 KB
• Volume: 4
• Category: Article
• ISSN: 0956-0521
• DOI: 10.1016/0956-0521(93)90018-r

✦ Synopsis

This paper explores the effectiveness of the author's decomposition in a parallel environment. The author's decomposition is a new class of direct method, which factorizes a nonsingular matrix into triangular matrices such that the product of the obtained triangular matrices is the inverse of the original matrix. This new technique provides an efficient procedure and a new feature in structural analysis. Both vectorization and parallelization are considered in this work. Performances in solving large-scaled sparse systems and comparisons with the well-known type of direct method, Gaussian elimination, are reported. Examples in application to domain decomposition are also included.