From: Christopher Creutzig <>
Newsgroups: comp.soft-sys.matlab
Subject: Re: Matrix Minimal Polynomial
Date: Wed, 09 Nov 2011 14:11:40 +0100
Organization: The MathWorks, Inc.
Lines: 18
Message-ID: <>
References: <is2omp$g31$> <is43mu$ph5$> <j9b5gj$ioq$>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
X-Trace: 1320844300 27335 (9 Nov 2011 13:11:40 GMT)
NNTP-Posting-Date: Wed, 9 Nov 2011 13:11:40 +0000 (UTC)
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:7.0.1) Gecko/20110929 Thunderbird/7.0.1
In-Reply-To: <j9b5gj$ioq$>
Xref: comp.soft-sys.matlab:748802

On 08.11.11 12:58, Pantelis S wrote:

> I guess MuPad uses symbolic algorithms to calculate the minimum polynomial - is there a function to do the same numerically? 

I don't think so. You probably could create something starting from
polyfun's poly.m (which computes the characteristic polynomial) by
checking the kernel before adding in another multiplicity of an
Eigenvalue, but off-hand I'm not sure if that is numerically stable.

> If not, is there a way to invoke linalg::minpoly from the workspace.

feval(symengine, 'expr', feval(symengine, 'linalg::minpoly', A, 'x'))

Kind of ugly, I know.