No label defined (Q4513041)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1526386
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 1526386 |
Statements
26 June 2001
0 references
parallel computing
0 references
spectral bisection
0 references
combinatorial graph theory
0 references
hybrid method
0 references
domain decomposition
0 references
numerical examples
0 references
finite element
0 references
Laplacian matrix
0 references
eigenvalue
0 references
eigenvector
0 references
Lanczos method
0 references
algorithm
0 references
0.813136100769043
0 references
0.8024395108222961
0 references
0.7829625010490417
0 references
0.7802831530570984
0 references
0.7802831530570984
0 references
A hybrid method for domain decomposition, employing combinatorial and algebraic graph theoretical algorithms, is presented. The approach uses the combinatorial graph theory for partial decomposition and then the spectral bisection approach. The spectral bisection is an algebraic graph method by which the associate graph of a finite element model is decomposed into two subgraphs. The optimality conditions consist of the equality of the numbers of nodes in the two subgraphs and minimality of the number of interface nodes. The Laplacian matrix is used to select a minimal cut set. The second eigenvalue and eigenvector of the Laplacian matrix are iteratively computed by the Lanczos method. For graph partitioning a hybrid method is proposed. In this algorithm the pseudo-peripheral nodes of the associate graph are first obtained. Four examples of partitioning illustrate the computational results.
0 references
