# HG changeset patch
# User Lachlan Andrew <lachlanbis@gmail.com>
# Date 1467275426 -36000
# Node ID 9a0e30e24b9bc943a92e33adbafe0c123e2d6b2a
# Parent  42456fc9bf6c0217fe855d64fc25b98fac86c32b
Calculate cov as Matlab does, not as its conjugate (bug #48315).

* NEWS: Announce change.

* cov.m: Remove "conj" from calculation of output.  Update compatibility note
to explain what Matlab actually does.

diff -r 42456fc9bf6c -r 9a0e30e24b9b NEWS
--- a/NEWS	Mon Aug 08 15:21:03 2016 -0700
+++ b/NEWS	Thu Jun 30 18:30:26 2016 +1000
@@ -1,8 +1,6 @@
 Summary of important user-visible changes for version 4.2:
 ---------------------------------------------------------
 
- ** condest now works with a normest1 compatible syntax.
-
  ** The parser has been extended to accept, but ignore, underscore
     characters in numbers.  This facilitates writing more legible code
     by using '_' as a thousands separator or to group nibbles into bytes
@@ -30,7 +28,7 @@
     to the old command "hold all" and was made for Matlab compatibility.
     Existing code *may* produce differently colored plots if it did not
     specify the color for a plot and relied on each new plot having the
-    the default first color in the "ColorOrder" property.
+    default first color in the "ColorOrder" property.
 
  ** When starting, Octave now looks in the function path for a file
     startup.m and executes any commands found there.  This change was
@@ -64,6 +62,12 @@
     compatible with Matlab releases newer than 2011.  In addition,
     Octave no longer supports matrix inputs for A or B.
 
+ ** The cov function now returns the complex conjugate of the result
+    from previous versions of Octave.  This change was made for
+    compatibility with Matlab.
+
+ ** condest now works with a normest1 compatible syntax.
+
  ** The griddata function no longer plots the interpolated mesh if no
     output argument is requested, instead the vector or array of
     interpolated values is always returned for Matlab compatibility.
diff -r 42456fc9bf6c -r 9a0e30e24b9b scripts/statistics/base/cov.m
--- a/scripts/statistics/base/cov.m	Mon Aug 08 15:21:03 2016 -0700
+++ b/scripts/statistics/base/cov.m	Thu Jun 30 18:30:26 2016 +1000
@@ -56,11 +56,13 @@
 ##   normalize with @math{N}, this provides the second moment around the mean
 ## @end table
 ##
-## Compatibility Note:: Octave always computes the covariance matrix.
-## For two inputs, however, @sc{matlab} will calculate
-## @code{cov (@var{x}(:), @var{y}(:))} whenever the number of elements in
-## @var{x} and @var{y} are equal.  This will result in a scalar rather than
-## a matrix output.  Code relying on this odd definition will need to be
+## Compatibility Note:: Octave always treats rows of @var{x} and @var{y}
+## as multivariate random variables.
+## For two inputs, however, @sc{matlab} treats @var{x} and @var{y} as two
+## univariate distributions regardless of their shapes, and will calculate
+## @code{cov ([@var{x}(:), @var{y}(:)])} whenever the number of elements in
+## @var{x} and @var{y} are equal.  This will result in a 2x2 matrix.
+## Code relying on @sc{Matlab}'s definition will need to be
 ## changed when running in Octave.
 ## @seealso{corr}
 ## @end deftypefn
@@ -108,7 +110,7 @@
 
   if (nargin == 1 || isscalar (y))
     x = center (x, 1);
-    c = conj (x' * x / (n - 1 + opt));
+    c = x' * x / (n - 1 + opt);
   else
     if (isrow (y))
       y = y.';
@@ -118,7 +120,7 @@
     endif
     x = center (x, 1);
     y = center (y, 1);
-    c = conj (x' * y / (n - 1 + opt));
+    c = x' * y / (n - 1 + opt);
   endif
 
 endfunction