Its OK now.
Thanks
Its OK now.
BTW, I have noticed another matlab compatibility issue with cpsd which
uses pwelch. It is probably matlab fault due to undocumented
inconsistency and not an octave fault.
I have noticed that matlab cpsd output for current versions (back at
least to 2006b) produce conjugate values in relation to older versions
such as ver 7.0.
I have not found any references for that inconsistency in any matlab
release notes. I did find some refernces in the internet from other
users that have faced this issue.
I have noticed that your pwelch function can handle compatibility with
matlab by setting a global variable but I am not sure if it does take
care for this undocumented inconsistent matlab behavior.
Anyway, thanks for your fast response.
Nitzan
Peter Lanspeary wrote:
On Fri, Mar 27, 2009 at 03:57:57AM +1030, Peter Lanspeary wrote:
Hi Nitzan,
Apologies. I'm responsible for that bug.
You have a good fix.
I'll take care of it tomorrow if it's not done before then.
Done.
On Thu, Mar 26, 2009 at 05:01:14PM +0200, nit ar wrote:
Hello octaveforge developers
I have found a bug in pwelch.m for the case when a cross power spectrum is
calculated and 'none' is passed as a parameter in order to remove mean from
both x and y inputs:
line 718:
elseif ( rm_mean ~= 2 ) % remove linear trend from segment
should be chamged to:
elseif ( rm_mean == 2 ) % remove linear trend from segment
Treatment of y should be equivalent to the treatment of x (line 707).
The above change produce results which are compatible with matlab.
Regards
Nitzan Arazi
------------------------------------------------------------------------------
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev
