Systolic computation of QZ matrix decomposition

DOI10.1080/00207169808804676MaRDI QIDQ4392894zbMATH OpenOpenAlexFDO

Authors D. J. Evans, Gabriel Okša

Publication date 20 December 1998

Published in International Journal of Computer Mathematics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1080/00207169808804676

zbMATH Keywords

Givens rotationHouseholder reflectionQR matrix decompositionQZ matrix decompositiontriangular systolic array

Mathematics Subject Classification ID

Direct numerical methods for linear systems and matrix inversion (65F05) Orthogonalization in numerical linear algebra (65F25)

Review text

QR and QZ matrix decompositions are efficient methods for the nullification of many matrix elements via orthogonal transformations. The authors describe at first traditional triangular array for QRD -- the point rotator and point reflector based on Givens rotation and Householder reflection, respectively. Four new triangular systolic arrays for QZD are described. Two of them are based on the block Givens transformation and other two arrays are based on the block Householder transformation. Their performance is compared with the point rotator and point reflector for the QR matrix decomposition. This comparison is based on the detailed discussion of the length of global time step which synchronizes the array, the number of required processors and the area defined by the vertical and horizontal links used in the array for the data pipelining. Some interesting observations have been made for the one from new implementations which carry out the block Givens rotation with an encoder and decoder in the diagonal and nondiagonal processors, respectively.

Recommendations

• A Rotation Method for Computing the QR-Decomposition
• Systolic Networks for Orthogonal Decompositions
• A size-independent systolic array for matrix triangularization and eigenvalue computation
• A systolic array architecture for QR decomposition of block structured sparse systems
• Systolic SVD and QR Decomposition by Householder Reflections

Cites work

• Block Reflectors: Theory and Computation
• scientific article; zbMATH DE number 4213315 (Why is no real title available?)
• Numerical linear algebra and applications
• Parallel householder method for linear systems
• Scaled Givens Rotations for the Solution of Linear Least Squares Problems on Systolic Arrays
• THE PARALLEL IMPLICIT ELIMINATION (PIE) METHOD FOR THE SOLUTION OF LINEAR SYSTEMS
• THE QZ ORTHOGONAL DECOMPOSITION METHOD

Cited in

(7)

• A systolic array architecture for QR decomposition of block structured sparse systems
• Systolic Networks for Orthogonal Decompositions
• Elimination of the computational broadcast in systolic arrays: an application to the qr decomposition algorithm
• Systolic SVD and QR Decomposition by Householder Reflections
• A Rotation Method for Computing the QR-Decomposition
• THE QZ ORTHOGONAL DECOMPOSITION METHOD
• Fast qz decomposition and its applications

This page was built for publication: Systolic computation of QZ matrix decomposition

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4392894)