Sexta-feira, 15 de outubro de 1999 Auditório Jacy Monteiro Instituto de Matemática e Estatística, USP 5:30 - 6:20 Marcos Kiwi, Universidad de Chile, Chile Min-Max-Boundary Domain Decomposition Domain decomposition is one of the most effective and popular parallel computing techniques for solving large scale numerical systems and partial differential equations numerically. It also has applications in circuit layout, simulations of the $N$-body problem, etc. When the amount of computation in a subdomain is proportional to the volume of the subdomain, domain decomposition amounts to minimizing the surface area of each subdomain while dividing the volume evenly. Motivated by this fact, we study the following min--max boundary multi--way partitioning problem: Given a graph $G$ and an integer $k$, we would like to divide $G$ into $k$ subgraphs (by removing edges) such that (i) the vertex size of each subgraph is $\Theta(|G|/k)$; and (ii) the maximum boundary size of any subgraph (the set of edges connecting it with other subgraphs) is minimized. We provide an algorithm that given a graph satisfying some hipotesis, in particular given a well--shaped mesh in $d$ dimensions, finds a partition of the graph into $k$ subgraphs, each one with $\Theta(|G|/k)$ vertices and the number of edges connecting each subgraph with the other subgraphs is $O((|G|/k)^{1-1/d})$. Finally, we extend our results to vertex--weighted and vertex--based graph decomposition. Our results can be used to simultaneously balance the computational and memory loads on a distributed--memory parallel computer without incurring large communication overhead. The results discussed in this talk are joint work with Dan Spielman (MIT) and Shang-Hua Teng (University of Illinois, Urbana-Champaign)
Last modified: Fri Oct 8 11:51:59 EDT 1999