Implementing the QR-algorithm on an array of processors
Thesis/Dissertation
·
OSTI ID:7106942
QR algorithms for solving the algebraic eigenvalue problem that initially reduce the matrix to upper Hessenberg form and utilize traditional shifting strategies do not lend themselves to efficient implementation on a grid of processors. This thesis introduces a variation of the QR algorithm that works with the full matrix and show how it can be implemented on a square array of processors. By using a deferred shifting scheme, iterations can be pipelined, thereby reducing processor idle time. A thorough analysis of deferred-shifting techniques show that the asymptotic convergence rate remains acceptable.
- Research Organization:
- Maryland Univ., College Park (USA)
- OSTI ID:
- 7106942
- Country of Publication:
- United States
- Language:
- English
Similar Records
New distributed-memory parallel algorithms for solving nonsymmetric eigenvalue problems
Regular iterative algorithms and their implementations on processor arrays
On a block implementation of Hessenberg multishift QR iteration: LAPACK working note No. 8
Conference
·
Thu Nov 30 23:00:00 EST 1995
·
OSTI ID:125574
Regular iterative algorithms and their implementations on processor arrays
Thesis/Dissertation
·
Tue Dec 31 23:00:00 EST 1985
·
OSTI ID:5310887
On a block implementation of Hessenberg multishift QR iteration: LAPACK working note No. 8
Technical Report
·
Sat Dec 31 23:00:00 EST 1988
·
OSTI ID:6524413