|
From: | Ismael Núñez-Riboni |
Subject: | Re: cov(x,y) = matrix or number? |
Date: | Mon, 12 Mar 2012 22:19:59 +0100 |
User-agent: | Mozilla/5.0 (X11; Linux i686; rv:10.0.2) Gecko/20120216 Thunderbird/10.0.2 |
Please refer to this discussion: http://octave.1599824.n4.nabble.com/Re-cov-m-patch-td4466691.html - Jordi G. H.
As I just wrote to the developer's list and after reading John's link I agree Octave's cov is improved in comparison to Matlab's cov (it seems to me that to do in Matlab what Octave does you need a loop), so I suggest not necessarily to patch the function but to handle with what is for many matlab users a counter intuitive behavior, i.e., either one of these:
1) Change Octave's function name, (perhaps cov2 or cov_i, for "improved").2) Insert a warning in the cov's help to drag attention to a different behavior than in Matlab.
If you want to make a version of cov compatible to Matlab I think my patch is not enough and this will degrade the advantages of Octave's cov function, i.e., it is not advisable... I take thus the patch suggestion back and rather suggest to modify cov's help as I wrote to the developers list... Saludos, Ismael.
[Prev in Thread] | Current Thread | [Next in Thread] |