[Octave-cvsupdate] SF.net SVN: octave: [5072] trunk/octave-forge/extra/tsa/inst/lpc.m

[<sch...@us...>]

Revision: 5072
          http://octave.svn.sourceforge.net/octave/?rev=5072&view=rev
Author:   schloegl
Date:     2008-05-20 01:19:53 -0700 (Tue, 20 May 2008)

Log Message:
-----------
lattice method fails for (some?) deterministic signals; make Levinson-Durbin the default method (thanks to Rafael Laboissiere for pointing this out

Modified Paths:
--------------
    trunk/octave-forge/extra/tsa/inst/lpc.m

Modified: trunk/octave-forge/extra/tsa/inst/lpc.m
===================================================================
--- trunk/octave-forge/extra/tsa/inst/lpc.m	2008-05-20 04:45:35 UTC (rev 5071)
+++ trunk/octave-forge/extra/tsa/inst/lpc.m	2008-05-20 08:19:53 UTC (rev 5072)
@@ -30,9 +30,10 @@
 %  M.B. Priestley "Spectral Analysis and Time Series" Academic Press, 1981. 
 %  W.S. Wei "Time Series Analysis" Addison Wesley, 1990.
 
-%	Version 2.91
-%	Copyright (C) 1996-2002 by Alois Schloegl <a.s...@ie...>
-%
+%	$Id$
+%	Copyright (C) 1996-2002,2008 by Alois Schloegl <a.s...@ie...>
+%       This is part of the TSA-toolbox. See also 
+%       http://hci.tugraz.at/schloegl/matlab/tsa/
 
 % This library is free software; you can redistribute it and/or
 % modify it under the terms of the GNU Library General Public
@@ -60,9 +61,10 @@
 % you can use any of the following routines. 
 % I've selected the Burg method, because it provides the most accurate estimates
 
-[AR,RC,PE] = lattice(Y.',P);		% Burg method
-% [AR,RC,PE] = lattice(Y.',P,'GEOL');	% geomatric lattice
-% [AR,RC,PE] = durlev(acovf(Y.',P));  	% Yule-Walker
 
+% [AR,RC,PE] = lattice(Y.',P);		% Burg method
+% [AR,RC,PE] = lattice(Y.',P,'GEOL');	% geometric lattice
+[AR,RC,PE] = durlev(acovf(Y.',P));  	% Yule-Walker
+
 A = ar2poly(AR);
 


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.