ENH: dpnp.histogram fallback; tests (#802)

* ENH: dpnp.histogram fallback; tests

Author: samir-nasibli

dpnp/dpnp_iface_statistics.py

@@ -54,6 +54,7 @@
     'bincount',
     'correlate',
     'cov',
+    'histogram',
     'max',
     'mean',
     'median',
@@ -295,6 +296,33 @@ def cov(x1, y=None, rowvar=True, bias=False, ddof=None, fweights=None, aweights=
     return call_origin(numpy.cov, x1, y, rowvar, bias, ddof, fweights, aweights)
 
 
+def histogram(a, bins=10, range=None, normed=None, weights=None, density=None):
+    """
+    Compute the histogram of a dataset.
+    For full documentation refer to :obj:`numpy.histogram`.
+    Examples
+    --------
+    >>> import dpnp
+    >>> dpnp.histogram([1, 2, 1], bins=[0, 1, 2, 3])
+    (array([0, 2, 1]), array([0, 1, 2, 3]))
+    >>> dpnp.histogram(dpnp.arange(4), bins=dpnp.arange(5), density=True)
+    (array([0.25, 0.25, 0.25, 0.25]), array([0, 1, 2, 3, 4]))
+    >>> dpnp.histogram([[1, 2, 1], [1, 0, 1]], bins=[0,1,2,3])
+    (array([1, 4, 1]), array([0, 1, 2, 3]))
+    >>> a = dpnp.arange(5)
+    >>> hist, bin_edges = dpnp.histogram(a, density=True)
+    >>> hist
+    array([0.5, 0. , 0.5, 0. , 0. , 0.5, 0. , 0.5, 0. , 0.5])
+    >>> hist.sum()
+    2.4999999999999996
+    >>> res = dpnp.sum(hist * dpnp.diff(bin_edges))
+    >>> print(res)
+    1.0
+    """
+
+    return call_origin(numpy.histogram, a=a, bins=bins, range=range, normed=normed, weights=weights, density=density)
+
+
 def max(x1, axis=None, out=None, keepdims=False, initial=None, where=True):
     """
     Return the maximum of an array or maximum along an axis.

tests/test_histograms.py

@@ -0,0 +1,73 @@
+import pytest
+
+import dpnp
+
+import numpy
+
+
+class TestHistogram:
+
+    def setup(self):
+        pass
+
+    def teardown(self):
+        pass
+
+    def test_simple(self):
+        n = 100
+        v = dpnp.random.rand(n)
+        (a, b) = dpnp.histogram(v)
+        # check if the sum of the bins equals the number of samples
+        numpy.testing.assert_equal(dpnp.sum(a, axis=0)[0], n)
+        # check that the bin counts are evenly spaced when the data is from
+        # a linear function
+        (a, b) = dpnp.histogram(numpy.linspace(0, 10, 100))
+        numpy.testing.assert_array_equal(a, 10)
+
+    def test_one_bin(self):
+        # Ticket 632
+        hist, edges = dpnp.histogram([1, 2, 3, 4], [1, 2])
+        numpy.testing.assert_array_equal(hist, [2, ])
+        numpy.testing.assert_array_equal(edges, [1, 2])
+        numpy.testing.assert_raises(ValueError, dpnp.histogram, [1, 2], bins=0)
+        h, e = dpnp.histogram([1, 2], bins=1)
+        numpy.testing.assert_equal(h, dpnp.array([2]))
+        numpy.testing.assert_allclose(e, dpnp.array([1., 2.]))
+
+    def test_density(self):
+        # Check that the integral of the density equals 1.
+        n = 100
+        v = dpnp.random.rand(n)
+        a, b = dpnp.histogram(v, density=True)
+        area = dpnp.sum(a * dpnp.diff(b)[0])[0]
+        numpy.testing.assert_almost_equal(area, 1)
+
+        # Check with non-constant bin widths
+        v = dpnp.arange(10)
+        bins = [0, 1, 3, 6, 10]
+        a, b = dpnp.histogram(v, bins, density=True)
+        numpy.testing.assert_array_equal(a, .1)
+        numpy.testing.assert_equal(dpnp.sum(a * dpnp.diff(b))[0], 1)
+
+        # Test that passing False works too
+        a, b = dpnp.histogram(v, bins, density=False)
+        numpy.testing.assert_array_equal(a, [1, 2, 3, 4])
+
+        # Variable bin widths are especially useful to deal with
+        # infinities.
+        v = dpnp.arange(10)
+        bins = [0, 1, 3, 6, numpy.inf]
+        a, b = dpnp.histogram(v, bins, density=True)
+        numpy.testing.assert_array_equal(a, [.1, .1, .1, 0.])
+
+        # Taken from a bug report from N. Becker on the numpy-discussion
+        # mailing list Aug. 6, 2010.
+        counts, dmy = dpnp.histogram(
+            [1, 2, 3, 4], [0.5, 1.5, numpy.inf], density=True)
+        numpy.testing.assert_equal(counts, [.25, 0])
+
+    def test_arr_weights_mismatch(self):
+        a = dpnp.arange(10) + .5
+        w = dpnp.arange(11) + .5
+        with numpy.testing.assert_raises_regex(ValueError, "same shape as"):
+            h, b = dpnp.histogram(a, range=[1, 9], weights=w, density=True)