Dynamic Matrices and Their Application to the Concept of a Semi-Sparse Matrix.
Abstract
This report primarily describes research performed by the author while he was still employed by the U.S. Army Chemical Research, Development and Engineering Center, as well as continuing research being conducted by the author at the U.S. Army Research Laboratory, Aberdeen, MD. In a number of cases (e.g.. tridiagonal matrices, triangular matrices, and the production of raster images from wireframe CAD data), one is left with the problem of efficiency storing and accessing matrices which are large enough to worry about, and have regions of a meaningful size with no data (or if one prefers, some default value, usually zero), and other regions with at least some data. In many cases, these matrices contain too much data to be efficiently handled using the linked lists commonly used with sparse matrices. On the other hand, they may be too large and too sparse to be efficiently handled using normal methods. The author has named the matrices which fall into this category, SEMI-SPARSE MATRICES. While it is true that there are well-known application-specific methods for dealing with some types of matrices in this category (e.g., triangular matrices), there appears to be no standard mend for dealing with this category as a whole. In response to some of the problems this caused, the author has developed and implemented the concept which he refers to as a DYNAMIC MATRIX.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1994
- Accession Number
- ADA288688
Entities
People
- Daniel M. Pressel
Organizations
- United States Army Research Laboratory