skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: New concurrent iterative methods with monotonic convergence

Abstract

This paper proposes the new concurrent iterative methods without using any derivatives for finding all zeros of polynomials simultaneously. The new methods are of monotonic convergence for both simple and multiple real-zeros of polynomials and are quadratically convergent. The corresponding accelerated concurrent iterative methods are obtained too. The new methods are good candidates for the application in solving symmetric eigenproblems.

Authors:
 [1]
  1. Michigan State Univ., East Lansing, MI (United States)
Publication Date:
Research Org.:
Front Range Scientific Computations, Inc., Lakewood, CO (United States)
OSTI Identifier:
433344
Report Number(s):
CONF-9604167-Vol.1
ON: DE96015306; TRN: 97:000720-0017
Resource Type:
Conference
Resource Relation:
Conference: Copper Mountain conference on iterative methods, Copper Mountain, CO (United States), 9-13 Apr 1996; Other Information: PBD: [1996]; Related Information: Is Part Of Copper Mountain conference on iterative methods: Proceedings: Volume 1; PB: 422 p.
Country of Publication:
United States
Language:
English
Subject:
99 MATHEMATICS, COMPUTERS, INFORMATION SCIENCE, MANAGEMENT, LAW, MISCELLANEOUS; ITERATIVE METHODS; CONVERGENCE; POLYNOMIALS; EIGENVALUES

Citation Formats

Yao, Qingchuan. New concurrent iterative methods with monotonic convergence. United States: N. p., 1996. Web.
Yao, Qingchuan. New concurrent iterative methods with monotonic convergence. United States.
Yao, Qingchuan. 1996. "New concurrent iterative methods with monotonic convergence". United States. doi:. https://www.osti.gov/servlets/purl/433344.
@article{osti_433344,
title = {New concurrent iterative methods with monotonic convergence},
author = {Yao, Qingchuan},
abstractNote = {This paper proposes the new concurrent iterative methods without using any derivatives for finding all zeros of polynomials simultaneously. The new methods are of monotonic convergence for both simple and multiple real-zeros of polynomials and are quadratically convergent. The corresponding accelerated concurrent iterative methods are obtained too. The new methods are good candidates for the application in solving symmetric eigenproblems.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = 1996,
month =
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • We have implemented a direct, multifrontal solution method with nested dissection ordering of finite element equations on a ten-processor ELXSI 6400 system (a shared memory architecture with local cache memory at each processor). A limitation of this concurrent direct method is an elimination tree that idles most processors while equations associated with the domain dissections are solved. A potential improvement in concurrent performance can be realized by replacing the elimination tree in the direct method with an embedded, highly concurrent, iterative solution of the dissection equations. Results feature the convergence and concurrent speedup behavior of the new method applied tomore » well-conditioned (Laplace equation) and poorly-conditioned (bending beam) problems in two dimensions.« less
  • What properties of a nonsymmetric matrix A determine the convergence rate of iterations such as GMRES, QMR, and Arnoldi? If A is far from normal, should one replace the usual Ritz values {r_arrow} eigenvalues notion of convergence of Arnoldi by alternative notions such as Arnoldi lemniscates {r_arrow} pseudospectra? Since Krylov subspace iterations can be interpreted as minimization processes involving polynomials of matrices, the answers to questions such as these depend upon mathematical problems of the following kind. Given a polynomial p(z), how can one bound the norm of p(A) in terms of (1) the size of p(z) on various setsmore » in the complex plane, and (2) the locations of the spectrum and pseudospectra of A? This talk reports some progress towards solving these problems. In particular, the authors present theorems that generalize the Kreiss matrix theorem from the unit disk (for the monomial A{sup n}) to a class of general complex domains (for polynomials p(A)).« less
  • The integral equations of magnetostatics, conventionally given in terms of the field variables M and H, are reformulated with M and B. Stability criteria and convergence rates of the eigenvectors of the linear iteration matrices are evaluated. The relaxation factor ..beta.. in the MH approach varies inversely with permeability ..mu.., and nonlinear problems with high permeability converge slowly. In contrast, MB iteration is stable for ..beta.. < 2, and nonlinear problems converge rapidly, at a rate essentially independent of ..mu... For a permeability of 10/sup 3/, the number of iterations is reduced by two orders of magnitude over the conventionalmore » method, and at higher permeabilities the reduction is proportionally greater. The dependence of MB convergence rate on ..beta.., degree of saturation, element aspect ratio, and problem size is found numerically. An analytical result for the MB convergence rate for small nonlinear problems is found to be accurate for ..beta..less than or equal to1.2. The results are generally valid for two- and three-dimensional integral methods and are independent of the particular discretization procedures used to compute the field matrix.« less
  • In [CDH] a method for the solution of boundary value problems for the biharmonic equation using conformal mapping was investigated. The method is an implementation of the classical method of Muskhelishvili. In [CDH] it was shown, using the Hankel structure, that the linear system in [Musk] is the discretization of the identify plus a compact operator, and therefore the conjugate gradient method will converge superlinearly. The purpose of this paper is to give an estimate of the superlinear convergence in the case when the boundary curve is in a Hoelder class.