Steps Towards an Iterative Rank-Revealing Method

ABSTRACT

This report describes advances towards the development of an iterative rank-revealing method. Our advances have been in developing a new row projection method that is similar to an LQ factorization. We refer to this as an orthogonal row projection method. We have also developed a new condition estimation scheme that is based on the orthogonal row projection method. Together it is expected that these components will lead to an iterative rank-revealing method. The motivation for this work comes from large sparse optimization and geometry problems. We examine one of these to help convey a more complete understanding of the relevance of this work and where it fits into industrial needs. The paper concludes with a list of problems, ideas, and research areas that we believe need to be pursued in order to produce an iterative rank-revealing method. This work was supported in part by National Science Foundation grants CCR-9413309 and DMS-9403224, and by Boeing Information and Support Services.

TECHNICAL REPORT

Boeing Information and Support Services, ISSTECH-95-013

November 30, 1995

CO-AUTHORS

Daniel Pierce

Carl D. Meyer

THE POSTSCRIPT FILE

The postscript file (uncompressed) for the entire paper is 812KB.

To receive it, click on RankRevealingMethod.ps

THE PDF FILE

To receive the PDF file, click on RankRevealingMethod.pdf

Return To Abstracts