Loading...
Please wait, while we are loading the content...
Similar Documents
Downdating a Rank-Revealing URV Decomposition
| Content Provider | Semantic Scholar |
|---|---|
| Author | Wu, Yuan-Jye Jason |
| Copyright Year | 1998 |
| Abstract | The rank-revealing URV decomposition is a useful tool for the subspace-tracking problem in digital signal processing. Updating the decomposition is a stable process. However, downdating a rank-revealing URV decomposition can be unstable because the R factor is ill-conditioned. In this article, we review some existing downdating algorithms for the full-rank URV decomposition in the absence of the U factor and develop a new combined algorithm. The combined algorithm has the merits of low cost and no intermediate breakdown, so the downdate is always computable in oating-point arithmetic. For the rank-revealing URV decomposition, we develop a two-step method that applies full-rank downdating algorithms to the signal and noise parts separately without using hyperbolic rotations. We prove that Park and Eld en's reduction algorithm and the combined algorithm have relational stabilities for both full-rank and rank-revealing cases. We demonstrate the eeciency and accuracy of our combined algorithm on ill-conditioned problems. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www-unix.mcs.anl.gov/~jwu/umcp/paper/downdate.ps |
| Alternate Webpage(s) | http://info.mcs.anl.gov/pub/tech_reports/reports/P658.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
