using rowvar flag in dpnp.cov

Author: Vahid Tavanashad

dpnp/dpnp_iface_statistics.py

@@ -44,6 +44,7 @@
 
 from dpnp.dpnp_algo import *
 from dpnp.dpnp_utils import *
+from dpnp.dpnp_array import dpnp_array
 import dpnp
 
 
@@ -237,7 +238,8 @@ def correlate(x1, x2, mode='valid'):
 
 
 def cov(x1, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=None):
-    """
+    """cov(m, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=None):
+
     Estimate a covariance matrix, given data and weights.
 
     For full documentation refer to :obj:`numpy.cov`.
@@ -248,7 +250,6 @@ def cov(x1, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=
     Dimension of input array ``m`` is limited by ``m.ndim > 2``.
     Size and shape of input arrays are supported to be equal.
     Prameters ``y`` is supported only with default value ``None``.
-    Prameters ``rowvar`` is supported only with default value ``True``.
     Prameters ``bias`` is supported only with default value ``False``.
     Prameters ``ddof`` is supported only with default value ``None``.
     Prameters ``fweights`` is supported only with default value ``None``.
@@ -280,8 +281,6 @@ def cov(x1, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=
             pass
         elif y is not None:
             pass
-        elif not rowvar:
-            pass
         elif bias:
             pass
         elif ddof is not None:
@@ -291,8 +290,14 @@ def cov(x1, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=
         elif aweights is not None:
             pass
         else:
-            if x1_desc.dtype != dpnp.float64:
-                x1_desc = dpnp.get_dpnp_descriptor(dpnp.astype(x1, dpnp.float64), copy_when_nondefault_queue=False)
+            if not rowvar and x1.shape[0] != 1:
+                x1 = x1.get_array() if isinstance(x1, dpnp_array) else x1
+                x1 = dpnp_array._create_from_usm_ndarray(x1.mT)
+                x1 = dpnp.astype(x1, dpnp.float64) if x1_desc.dtype != dpnp.float64 else x1
+                x1_desc = dpnp.get_dpnp_descriptor(x1, copy_when_nondefault_queue=False)
+            elif x1_desc.dtype != dpnp.float64:
+                x1 = dpnp.astype(x1, dpnp.float64)
+                x1_desc = dpnp.get_dpnp_descriptor(x1, copy_when_nondefault_queue=False)
 
             return dpnp_cov(x1_desc).get_pyobj()
 

tests/test_statistics.py

@@ -1,5 +1,5 @@
 import pytest
-
+from .helper import get_all_dtypes
 import dpnp
 
 import numpy
@@ -114,3 +114,18 @@ def test_bincount_weights(self, array, weights):
         expected = numpy.bincount(np_a, weights=weights)
         result = dpnp.bincount(dpnp_a, weights=weights)
         numpy.testing.assert_array_equal(expected, result)
+
+@pytest.mark.parametrize("dtype", get_all_dtypes(no_bool=True, no_none=True, no_complex=True))
+def test_cov_rowvar1(dtype):
+    a = dpnp.array([[0, 2], [1, 1], [2, 0]], dtype=dtype)
+    b = numpy.array([[0, 2], [1, 1], [2, 0]], dtype=dtype)
+    numpy.testing.assert_array_equal(dpnp.cov(a.T), dpnp.cov(a,rowvar=False))
+    numpy.testing.assert_array_equal(numpy.cov(b,rowvar=False), dpnp.cov(a,rowvar=False))
+
+@pytest.mark.parametrize("dtype", get_all_dtypes(no_bool=True, no_none=True, no_complex=True))
+def test_cov_rowvar2(dtype):
+    a = dpnp.array([[0, 1, 2]], dtype=dtype)
+    b = numpy.array([[0, 1, 2]], dtype=dtype)
+    numpy.testing.assert_array_equal(numpy.cov(b,rowvar=False), dpnp.cov(a,rowvar=False))
+
+