An Efficient and Stable Parallel Solution for Non-symmetric Toeplitz Linear Systems
نویسندگان
چکیده
In this paper, we parallelize a new algorithm for solving non– symmetric Toeplitz linear systems. This algorithm embeds the Toeplitz matrix in a larger structured matrix, then transforms it into an embedded Cauchy–like matrix by means of trigonometric modifications. Finally, the algorithm applies a modified QR transformation to triangularize the augmented matrix. The algorithm combines efficiency and stability. It has been implemented using standard tools and libraries, thereby producing a portable code. An extensive experimental analysis has been performed on a cluster of personal computers. Experimental results show that we can obtain efficiencies that are similar to other fast parallel algorithms, while obtaining more accurate results with only one iterative refinement step in the solution.
منابع مشابه
Parallelizing the Conjugate Gradient Algorithm for Multilevel Toeplitz Systems
Multilevel Toeplitz linear systems appear in a wide range of scientific and engineering applications. While several fast direct solvers exist for the basic 1-level Toeplitz matrices, in the multilevel case an iterative solver provides the most general and practical solution. Furthermore, iterative methods are asymptotically faster than many stable direct methods even for the 1-level case. This ...
متن کاملParallel Algorithms for the Solution of Toeplitz Systems of Linear Equations
In this paper we present two parallel algorithms to solve non-symmetric Toeplitz systems of linear equations. The first algorithm performs a modified QR factorization of the matrix by using the generalized Schur algorithm. The second one is based on the transformation of the Toeplitz matrix into a Cauchy-like matrix in order to reduce the communication cost. Both sequential methods have small c...
متن کاملParallel Solution of Toeplitz and Toeplitz-Like Linear Systems Over Fields of Small Positive Characteristic
We show that over a field of characteristic p the solution to a non-singular system of n linear equations in n unknows, with 2 ≤ p < n, whose coefficient matrix is of displacement rank α and which is given as a sum of α LU-products of Toeplitz matrices, can be computed in parallel with randomization simultaneously in O((log n)) time and a total work of O(max{αn, p}n× log n loglog n). A time uni...
متن کاملPreconditioners for ill { conditioned Toeplitz matrices
This paper is concerned with the solution of systems of linear equations ANx = b, where fANg N2N denotes a sequence of positive deenite Hermitian ill{conditioned Toeplitz matrices arising from a (real{valued) nonnegative generating function f 2 C2 with zeros. We construct positive deenite Hermitian preconditioners MN such that the eigenvalues of M ?1 N AN are clustered at 1 and the correspondin...
متن کاملP-stability, TF and VSDPL technique in Obrechkoff methods for the numerical solution of the Schrodinger equation
Many simulation algorithms (chemical reaction systems, differential systems arising from the modeling of transient behavior in the process industries and etc.) contain the numerical solution of systems of differential equations. For the efficient solution of the above mentioned problems, linear multistep methods or Runge-Kutta technique are used. For the simulation of chemical procedures the ra...
متن کامل