Naslov
Accelerating Block Jacobi Methods

Autori
Hari, Vjeran

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, neobjavljeni rad, znanstveni

Skup
International GAMM Workshop on Guaranteed Error-Bounds for the Solution of Nonlinear Problems in Applied Mathematics

Mjesto i datum
München, Njemačka, 01.09.2003. - 05.09.2003

Vrsta sudjelovanja
Pozvano predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Block Jacobi Methods; Accuracy; Efficiency

Sažetak
Until recently QR methods have been considered as the fastest for computing eigenvalue and singular value problems. Lately, Divide and Conquor ethods have been developed to a stage that overcomes the performance of the QR methods. However, the both types of methods, require reducing the initial full matrix to tridiagonal or bidiagonal form. This preprocessing is dangerous since it can deteriorate accuracy of smaller eigenvalues or singular values. In addition, reducing the matrix to tridiagonal or bidiagonal form is costly:the later part of QR or DC methods has been improving One-sided Jacobi-type methods for solving eigenvalue and singular value problems are known for their efficiency on ordinary and parallel computers and for their relative accuracy. In order to further enhance efficiency, block versions of these methods can be designed. It is shown, on the example of one-sided block-Jacobi method for computing the singular value decomposition, how CS decomposition and some other tricks can be used to reduce the flop count in the slowest part of the algorithm -- updating the block columns.

Izvorni jezik
Engleski

POVEZANOST RADA