Recursive Mesh Refinement on Hypercubes
Abstract
Adaptive methods for partial differential equations can be viewed as a graph problem. Parallel methods must distribute this graph efficiently among the processors. In doing this, the cost of communication between processors and the structure of the graph must be considered. The author divides this problem into two phases: labeling of graph nodes and subsequent mapping of these labels onto processors. He describes a new form of Gray-code which we calls an interleaved Gray-code that allows easy labeling of graph nodes even when the maximal level of refinement is unknown, allows easy determination of nearby nodes in the graph, is completely deterministic, and often (in a well-defined sense) distributes the graph efficiently across a hypercube. The theoretical results are supported by computational experiments on the Connection Machine.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1988
- Accession Number
- ADA198695
Entities
People
- I. C. Ipsen
- W. D. Gropp
Organizations
- Yale University