From: Eric Durant <edurant@engin.umich.edu>
Subject: Re: modulo2 matrix inversion
Date: 09 Mar 2000 00:00:00 GMT
Message-ID: <38C866F9.A7108FC@engin.umich.edu>
Content-Transfer-Encoding: 7bit
References: <38BAA81B.A38DA84C@Infineon.com> <38C41A2A.FC237F11@umich.edu>
X-Accept-Language: en,pdf
Content-Type: text/plain; charset=us-ascii
X-Trace: news.itd.umich.edu 952657651 141.213.6.34 (Thu, 09 Mar 2000 22:07:31 EST)
MIME-Version: 1.0
NNTP-Posting-Date: Thu, 09 Mar 2000 22:07:31 EST
Newsgroups: comp.soft-sys.matlab

Eric Durant wrote:
[...]
> For example, if your matrix to invert is in A_GF, the following 
> gives the inverse over GF(2) in AI_GF.  This does not check that 
> the matrix is invertible over the field.
> 
> p = 2; % Field characteristic, must be prime for this method

Whoops.  Of course, the characteristic is always prime.  I meant that
the field size must be prime for this method.

> AI_Q = inv(A_GF);

Better yet, to get the rationals without fooling around with the
floating point representation:

AI_Q_sym = inv(sym(A_GF))

And then proceed as in the previous post (in concept -- different code
of course to work with the symbolic representations).

I suspect there is an efficient way to do this by calling Maple; Maple
has powerful finite field functions, but I'm not familiar with its
syntax.  Perhaps some variation on the following specifying that the
field to operate in is GF(2), not Q(Z)...

AI_GF = maple('inv',sym(A_GF))

-- Eric Durant
   http://www.edurant.com/