Parallel algorithms for linear recurrences and for matrix and vector computations.

Title: Parallel algorithms for linear recurrences and for matrix and vector computations.
Identifier: AAI9530851
identifier: 9530851
Creator: Atinkpahoun, Antoine Marcel.
Contributor: Adviser: Victor Y. Pan
Date: 1995
Language: English
Publisher: City University of New York.
Subject: Computer Science | Mathematics | Geophysics
Abstract: The purposes of this study were: (1) demonstrate close correlations of banded matrix computations and computation of linear recurrence sequences, and also correlation among several known divide-and-conquer algorithms, that is, nested dissection, block cyclic reduction and prefix computation, (2) present some recent algorithms for computations with banded matrices and (3) modify the known FFT based algorithm for multiplying two polynomials or, equivalently, for computing the convolution of two vectors (of their coefficients), so as to decrease the requirement to the storage space used at the expense of a relatively small slowdown of the computation.
Type: dissertation
Source: PQT Legacy CUNY.xlsx
degree: Ph.D.

Media: Parallel algorithms for linear recurrences and for matrix and vector computations.