Newsgroups: comp.parallel
From: alberto@moreira.MV.COM (Alberto C Moreira)
Subject: Re: [Q]: Distributed eigenvector/eigenvalue
Keywords: eigenvector
Organization: MV Communications, Inc.
Date: Thu, 23 Nov 1995 22:44:11 GMT
Message-ID: <alberto.548.0011BD07@moreira.mv.com>

In article <48dblo$h8d@scapa.cs.ualberta.ca> rand@cs.ualberta.ca (Randal Kornelsen) writes:

>I am looking at implementing an algorithm for distributed 
>eigenvector/value computation.  Many of the algorithms that I
>have looked at are iterative in nature (eg. Jacobi).   Are there
>better algorithms to use as the starting point?   Are there algorithms
>and/or code that is publicly available?

   Finding the Eigenvalues of a matrix is equivalent to finding all roots of a 
   polinomial. Niels Abel proved that there can be no formula for solving
   polinomial equations of degree greater than four.

   In other words, methods to computer Eigenvalues must be iterative. 
   You can check "Numerical Recipes" for code that works, or for example,
   
        "Fundamentals of Matrix Computations",David S. Watkins, 
        John Wiley, 1991.

   in cas you'd rather write your own code. Hope this helps,


                                                                    _alberto_

           


