|
From: | David Bateman |
Subject: | Re: rank(gf) doesn't really work? |
Date: | Tue, 11 Oct 2005 15:51:15 +0200 |
User-agent: | Mozilla Thunderbird 0.8 (X11/20040923) |
Lapo Luchini wrote:
Yes this is a bug.. The rank function does an LU factorization of the galois array which pivots where it can and then attempts to count the rank deficiencies where it can't. Unfortunately, I messed this up. The relevant function is galois.cc (LU::factor) and the "info" variable. I'll look at it and try and fix it...Maybe I'm missing some important point (e.g. "gfrank" is not defined) and this is really not supported, but it gives different result on gf() than on a normal array so just maybe it shuold work..? $ B=gf([0,1,0,0,1,1; 0,0,1,1,1,0; 0,1,1,1,0,0; 1,0,1,1,0,0; 1,1,0,1,0,0; 1,1,1,0,0,0]); $ D=gf([0,1,0,0,1,1; 0,0,1,1,1,0; 0,1,1,1,0,0; 1,0,1,1,0,0; 1,1,0,1,0,0; 0,1,1,0,0,0]); $ B+D ans = GF(2) array. Array elements = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 $ rank(B) ans = 6 $ rank(D) ans = 4 Then, if B is full-rank (6 out of 6) how can possibly D be rank 4 if the first 5 lines are the very same 5 lines that in B were linearly independent? It just seems that it works, but only on square matrices, and it just ignores any other column: $ rank(B(1:5,1:5)) ans = 4 $ rank(B(1:5,2:6)) ans = 5 $ rank(B(1:5,:)) ans = 4 I would expect last line to be 5 as well. Maybe rank(B(1:5,:)) just uses the first 5 columns in the 5x6 matrix? Mhh...? Lapo
D. -- David Bateman address@hiddenMotorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 1 69 35 77 01 (Fax) 91193 Gif-Sur-Yvette FRANCE
The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary
------------------------------------------------------------- Octave is freely available under the terms of the GNU GPL. Octave's home on the web: http://www.octave.org How to fund new projects: http://www.octave.org/funding.html Subscription information: http://www.octave.org/archive.html -------------------------------------------------------------
[Prev in Thread] | Current Thread | [Next in Thread] |