A Parallel Qr-factorization/solver of Quasiseparable Matrices
نویسندگان
چکیده
Abstract. This manuscript focuses on the development of a parallel QR-factorization of structured rank matrices, which can then be used for solving systems of equations. First, we will prove the existence of two types of Givens transformations, named rank decreasing and rank expanding Givens transformations. Combining these two types of Givens transformations leads to different patterns for annihilating the lower triangular part of structured rank matrices. How to obtain different annihilation patterns, for computing the upper triangular factor R, such as the and pattern will be investigated. Another pattern, namely the -pattern, will be used for computing the QR-factorization in a parallel way. As an example of such a parallel QR-factorization, we will implement it for a quasiseparable matrix. This factorization can be run on 2 processors, with one step of intermediate communication in which one row needs to be sent from one processor to the other and back. Another example, showing how to deduce a parallel QR-factorization for a more general rank structure will also be discussed. Numerical experiments are included for demonstrating the accuracy and speed of this parallel algorithm w.r.t. the existing factorization of quasiseparable matrices. Also some numerical experiments on solving systems of equations using this approach will be given.
منابع مشابه
Stability of QR-based fast system solvers for a subclass of quasiseparable rank one matrices
The development of fast algorithms to perform computations with quasiseparable matrices has received a lot of attention in the last decade. Many different algorithms have been presented by several research groups all over the world. Despite this intense activity, to the best of our knowledge, there is no rounding error analysis published for these fast algorithms. In this paper, we present erro...
متن کاملAlgorithm 9xx: Sparse QR Factorization on the GPU
Sparse matrix factorization involves a mix of regular and irregular computation, which is a particular challenge when trying to obtain high-performance on the highly parallel general-purpose computing cores available on graphics processing units (GPUs). We present a sparse multifrontal QR factorization method that meets this challenge, and is up to eleven times faster than a highly optimized me...
متن کاملThe QR iteration method for Hermitian quasiseparable matrices of an arbitrary order
The QR iteration method for tridiagonal matrices is in the heart of one classical method to solve the general eigenvalue problem. In this paper we consider the more general class of quasiseparable matrices that includes not only tridiagonal but also companion, comrade, unitary Hessenberg and semiseparble matrices. A fast QR iteration method exploiting the Hermitian quasiseparable structure (and...
متن کاملParallel tools for solving incremental dense least squares
We present a parallel distributed solver that enables us to solve incremental dense least squares arising in some parameter estimation problems. This solver is based on ScaLAPACK [8] and PBLAS [9] kernel routines. In the incremental process, the observations are collected periodically and the solver updates the solution with new observations using a QR factorization algorithm. It uses a recentl...
متن کاملParallel tools for solving incremental dense least squares problems: application to space geodesy
We present a parallel distributed solver that enables us to solve incremental dense least squares arising in some parameter estimation problems. This solver is based on ScaLAPACK [8] and PBLAS [9] kernel routines. In the incremental process, the observations are collected periodically and the solver updates the solution with new observations using a QR factorization algorithm. It uses a recentl...
متن کامل