Wed 10 Feb 2016 10:05:00 PM UTC, comment #9:
Marking as "Won't Fix" and closing report.
|
Wed 10 Feb 2016 03:48:55 AM UTC, comment #8:
Rik, thanks for checking. Matlab now also fails on matrices greater than two dimensions. (I mistakenly said "more than three" last time.)
This is no longer a compatibility issue. I'll mark it as a feature request, but think it should be closed -- Matlab moved away from providing that feature.
|
Thu 08 Oct 2015 12:35:14 AM UTC, comment #7:
I think this Matlab compatibility request is still present. If I try the code from comment #4 in Octave dev (2015/10/07) I get
|
Thu 17 Sep 2015 09:27:27 AM UTC, comment #6:
I believe corrcoef has been removed from Octave 4.0.0.
Also, corr produces an error in both Matlab and Octave for arrays of more than 3 dimensions, and corrcoef does in Matlab.
Can this bug be closed now?
|
Sun 28 Jul 2013 03:44:55 PM UTC, comment #5:
If I understand correctly where this left off, this is still a bug in the development version.
|
Sun 17 Jul 2011 04:10:54 AM UTC, comment #4:
Rik,
This looks like what you want:
>> corr (rand(2,2,3), rand(2,2,3))
ans =
-1.0000 -1.0000
-1.0000 -1.0000
1.0000 1.0000
-1.0000 -1.0000
-1.0000 -1.0000
-1.0000 -1.0000
>>
|
Sun 17 Jul 2011 03:46:54 AM UTC, comment #3:
If corr only works on 2-D arrays then I think our current implementation is fine and I can close this report.
Perhaps, one could try
which would at least show if the NaN output in the previous example was caused by not having a true matrix in the first two dimensions.
|
Sun 17 Jul 2011 03:14:55 AM UTC, comment #2:
>> corr (rand(1,1,10), rand(1,1,10))
ans =
NaN
NaN
NaN
NaN
NaN
NaN
NaN
NaN
NaN
NaN
>>
As usual, this is Matlab r2009b. I tried a few other things
with corr(), but I need to reread its Matlab decription...
|
Sun 17 Jul 2011 02:05:56 AM UTC, comment #1:
Since the function corrcoef in Octave is more equivalent to the function corr() in Matlab, could you report whether the following code produces an error?
|
Sat 11 Jun 2011 03:50:48 PM UTC, original submission:
corrcoef(rand(1,1,10), rand(1,1,10))
% crashes under Octave but works in Matlab
|