Implement dpnp.isneginf and dpnp.isposinf (#1888)

* Implement dpnp.isneginf()

* Add tests for dpnp.isneginf()

* Implement dpnp.isposinf()

* Add tests for dpnp.isposinf()

* Add new functions to gen docs

* Add additional checks

* Add test_infinity_sign_errors

* Add sycl_queue/usm tests for logic functions

* Update tests

* Remove out dtype check

* Add TODO with support different out dtype

* Update test_logic_op_2in

---------

Co-authored-by: Anton <[email protected]>

Author: vlad-perevezentsev

doc/reference/logic.rst

@@ -26,6 +26,8 @@ Infinities and NaNs
    dpnp.isfinite
    dpnp.isinf
    dpnp.isnan
+   dpnp.isneginf
+   dpnp.isposinf
 
 
 Array type testing

dpnp/dpnp_iface_logic.py

@@ -66,6 +66,8 @@
     "isfinite",
     "isinf",
     "isnan",
+    "isneginf",
+    "isposinf",
     "less",
     "less_equal",
     "logical_and",
@@ -777,6 +779,150 @@ def isclose(x1, x2, rtol=1e-05, atol=1e-08, equal_nan=False):
 )
 
 
+def isneginf(x, out=None):
+    """
+    Test element-wise for negative infinity, return result as bool array.
+
+    For full documentation refer to :obj:`numpy.isneginf`.
+
+    Parameters
+    ----------
+    x : {dpnp.ndarray, usm_ndarray}
+        Input array.
+    out : {None, dpnp.ndarray, usm_ndarray}, optional
+        A location into which the result is stored. If provided, it must have a
+        shape that the input broadcasts to and a boolean data type.
+        If not provided or ``None``, a freshly-allocated boolean array
+        is returned.
+        Default: ``None``.
+
+    Returns
+    -------
+    out : dpnp.ndarray
+        Boolean array of same shape as ``x``.
+
+    See Also
+    --------
+    :obj:`dpnp.isinf` : Test element-wise for positive or negative infinity.
+    :obj:`dpnp.isposinf` : Test element-wise for positive infinity,
+                            return result as bool array.
+    :obj:`dpnp.isnan` : Test element-wise for NaN and
+                    return result as a boolean array.
+    :obj:`dpnp.isfinite` : Test element-wise for finiteness.
+
+    Examples
+    --------
+    >>> import dpnp as np
+    >>> x = np.array(np.inf)
+    >>> np.isneginf(-x)
+    array(True)
+    >>> np.isneginf(x)
+    array(False)
+
+    >>> x = np.array([-np.inf, 0., np.inf])
+    >>> np.isneginf(x)
+    array([ True, False, False])
+
+    >>> x = np.array([-np.inf, 0., np.inf])
+    >>> y = np.zeros(x.shape, dtype='bool')
+    >>> np.isneginf(x, y)
+    array([ True, False, False])
+    >>> y
+    array([ True, False, False])
+
+    """
+
+    dpnp.check_supported_arrays_type(x)
+
+    if out is not None:
+        dpnp.check_supported_arrays_type(out)
+
+    x_dtype = x.dtype
+    if dpnp.issubdtype(x_dtype, dpnp.complexfloating):
+        raise TypeError(
+            f"This operation is not supported for {x_dtype} values "
+            "because it would be ambiguous."
+        )
+
+    is_inf = dpnp.isinf(x)
+    signbit = dpnp.signbit(x)
+
+    # TODO: support different out dtype #1717(dpctl)
+    return dpnp.logical_and(is_inf, signbit, out=out)
+
+
+def isposinf(x, out=None):
+    """
+    Test element-wise for positive infinity, return result as bool array.
+
+    For full documentation refer to :obj:`numpy.isposinf`.
+
+    Parameters
+    ----------
+    x : {dpnp.ndarray, usm_ndarray}
+        Input array.
+    out : {None, dpnp.ndarray, usm_ndarray}, optional
+        A location into which the result is stored. If provided, it must have a
+        shape that the input broadcasts to and a boolean data type.
+        If not provided or ``None``, a freshly-allocated boolean array
+        is returned.
+        Default: ``None``.
+
+    Returns
+    -------
+    out : dpnp.ndarray
+        Boolean array of same shape as ``x``.
+
+    See Also
+    --------
+    :obj:`dpnp.isinf` : Test element-wise for positive or negative infinity.
+    :obj:`dpnp.isneginf` : Test element-wise for negative infinity,
+                            return result as bool array.
+    :obj:`dpnp.isnan` : Test element-wise for NaN and
+                    return result as a boolean array.
+    :obj:`dpnp.isfinite` : Test element-wise for finiteness.
+
+    Examples
+    --------
+    >>> import dpnp as np
+    >>> x = np.array(np.inf)
+    >>> np.isposinf(x)
+    array(True)
+    >>> np.isposinf(-x)
+    array(False)
+
+    >>> x = np.array([-np.inf, 0., np.inf])
+    >>> np.isposinf(x)
+    array([False, False,  True])
+
+    >>> x = np.array([-np.inf, 0., np.inf])
+    >>> y = np.zeros(x.shape, dtype='bool')
+    >>> np.isposinf(x, y)
+    array([False, False,  True])
+    >>> y
+    array([False, False,  True])
+
+    """
+
+    dpnp.check_supported_arrays_type(x)
+
+    if out is not None:
+        dpnp.check_supported_arrays_type(out)
+
+    x_dtype = x.dtype
+    if dpnp.issubdtype(x_dtype, dpnp.complexfloating):
+        raise TypeError(
+            f"This operation is not supported for {x_dtype} values "
+            "because it would be ambiguous."
+        )
+
+    is_inf = dpnp.isinf(x)
+    signbit = ~dpnp.signbit(x)
+
+    # TODO: support different out dtype #1717(dpctl)
+    return dpnp.logical_and(is_inf, signbit, out=out)
+
+
 _LESS_DOCSTRING = """
 Computes the less-than test results for each element `x1_i` of
 the input array `x1` with the respective element `x2_i` of the input array `x2`.

tests/test_logic.py

@@ -7,6 +7,7 @@
 from .helper import (
     get_all_dtypes,
     get_float_complex_dtypes,
+    get_float_dtypes,
 )
 
 
@@ -432,3 +433,49 @@ def test_finite(op, data, dtype):
     dpnp_res = getattr(dpnp, op)(x, out=dp_out)
     assert dp_out is dpnp_res
     assert_equal(dpnp_res, np_res)
+
+
+@pytest.mark.parametrize("func", ["isneginf", "isposinf"])
+@pytest.mark.parametrize(
+    "data",
+    [
+        [dpnp.inf, -1, 0, 1, dpnp.nan, -dpnp.inf],
+        [[dpnp.inf, dpnp.nan], [dpnp.nan, 0], [1, -dpnp.inf]],
+    ],
+    ids=[
+        "1D array",
+        "2D array",
+    ],
+)
+@pytest.mark.parametrize("dtype", get_float_dtypes())
+def test_infinity_sign(func, data, dtype):
+    x = dpnp.asarray(data, dtype=dtype)
+    np_res = getattr(numpy, func)(x.asnumpy())
+    dpnp_res = getattr(dpnp, func)(x)
+    assert_equal(dpnp_res, np_res)
+
+    dp_out = dpnp.empty(np_res.shape, dtype=dpnp.bool)
+    dpnp_res = getattr(dpnp, func)(x, out=dp_out)
+    assert dp_out is dpnp_res
+    assert_equal(dpnp_res, np_res)
+
+
+@pytest.mark.parametrize("func", ["isneginf", "isposinf"])
+def test_infinity_sign_errors(func):
+    data = [dpnp.inf, 0, -dpnp.inf]
+
+    # unsupported data type
+    x = dpnp.asarray(data, dtype="c8")
+    x_np = dpnp.asnumpy(x)
+    assert_raises(TypeError, getattr(dpnp, func), x)
+    assert_raises(TypeError, getattr(numpy, func), x_np)
+
+    # unsupported type
+    assert_raises(TypeError, getattr(dpnp, func), data)
+    assert_raises(TypeError, getattr(dpnp, func), x_np)
+
+    # unsupported `out` data type
+    x = dpnp.asarray(data, dtype=dpnp.default_float_type())
+    out = dpnp.empty_like(x, dtype="int32")
+    with pytest.raises(ValueError):
+        getattr(dpnp, func)(x, out=out)

tests/test_sycl_queue.py

@@ -501,6 +501,40 @@ def test_1in_1out(func, data, device):
     assert_sycl_queue_equal(result_queue, expected_queue)
 
 
+@pytest.mark.parametrize(
+    "op",
+    [
+        "all",
+        "any",
+        "isfinite",
+        "isinf",
+        "isnan",
+        "isneginf",
+        "isposinf",
+        "logical_not",
+    ],
+)
+@pytest.mark.parametrize(
+    "device",
+    valid_devices,
+    ids=[device.filter_string for device in valid_devices],
+)
+def test_logic_op_1in(op, device):
+    x = dpnp.array(
+        [-dpnp.inf, -1.0, 0.0, 1.0, dpnp.inf, dpnp.nan], device=device
+    )
+    result = getattr(dpnp, op)(x)
+
+    x_orig = dpnp.asnumpy(x)
+    expected = getattr(numpy, op)(x_orig)
+    assert_dtype_allclose(result, expected)
+
+    expected_queue = x.get_array().sycl_queue
+    result_queue = result.get_array().sycl_queue
+
+    assert_sycl_queue_equal(result_queue, expected_queue)
+
+
 @pytest.mark.parametrize(
     "device",
     valid_devices,
@@ -705,6 +739,55 @@ def test_2in_1out(func, data1, data2, device):
     assert_sycl_queue_equal(result.sycl_queue, x2.sycl_queue)
 
 
+@pytest.mark.parametrize(
+    "op",
+    [
+        "equal",
+        "greater",
+        "greater_equal",
+        # TODO: unblock when dpnp.isclose() is updated
+        # "isclose",
+        "less",
+        "less_equal",
+        "logical_and",
+        "logical_or",
+        "logical_xor",
+        "not_equal",
+    ],
+)
+@pytest.mark.parametrize(
+    "device",
+    valid_devices,
+    ids=[device.filter_string for device in valid_devices],
+)
+def test_logic_op_2in(op, device):
+    x1 = dpnp.array(
+        [-dpnp.inf, -1.0, 0.0, 1.0, dpnp.inf, dpnp.nan], device=device
+    )
+    x2 = dpnp.array(
+        [dpnp.inf, 1.0, 0.0, -1.0, -dpnp.inf, dpnp.nan], device=device
+    )
+    # Remove NaN value from input arrays because numpy raises RuntimeWarning
+    if op in [
+        "greater",
+        "greater_equal",
+        "less",
+        "less_equal",
+    ]:
+        x1 = x1[:-1]
+        x2 = x2[:-1]
+    result = getattr(dpnp, op)(x1, x2)
+
+    x1_orig = dpnp.asnumpy(x1)
+    x2_orig = dpnp.asnumpy(x2)
+    expected = getattr(numpy, op)(x1_orig, x2_orig)
+
+    assert_dtype_allclose(result, expected)
+
+    assert_sycl_queue_equal(result.sycl_queue, x1.sycl_queue)
+    assert_sycl_queue_equal(result.sycl_queue, x2.sycl_queue)
+
+
 @pytest.mark.parametrize(
     "func, data, scalar",
     [

tests/test_usm_type.py

@@ -357,20 +357,32 @@ def test_tril_triu(func, usm_type):
 @pytest.mark.parametrize(
     "op",
     [
-        "equal",
-        "greater",
-        "greater_equal",
-        "less",
-        "less_equal",
-        "logical_and",
-        "logical_or",
-        "logical_xor",
-        "not_equal",
+        "all",
+        "any",
+        "isfinite",
+        "isinf",
+        "isnan",
+        "isneginf",
+        "isposinf",
+        "logical_not",
     ],
-    ids=[
+)
+@pytest.mark.parametrize("usm_type_x", list_of_usm_types, ids=list_of_usm_types)
+def test_coerced_usm_types_logic_op_1in(op, usm_type_x):
+    x = dp.arange(-10, 10, usm_type=usm_type_x)
+    res = getattr(dp, op)(x)
+
+    assert x.usm_type == res.usm_type == usm_type_x
+
+
+@pytest.mark.parametrize(
+    "op",
+    [
         "equal",
         "greater",
         "greater_equal",
+        # TODO: unblock when dpnp.isclose() is updated
+        # "isclose",
         "less",
         "less_equal",
         "logical_and",
@@ -381,7 +393,7 @@ def test_tril_triu(func, usm_type):
 )
 @pytest.mark.parametrize("usm_type_x", list_of_usm_types, ids=list_of_usm_types)
 @pytest.mark.parametrize("usm_type_y", list_of_usm_types, ids=list_of_usm_types)
-def test_coerced_usm_types_logic_op(op, usm_type_x, usm_type_y):
+def test_coerced_usm_types_logic_op_2in(op, usm_type_x, usm_type_y):
     x = dp.arange(100, usm_type=usm_type_x)
     y = dp.arange(100, usm_type=usm_type_y)[::-1]
 

tests/third_party/cupy/logic_tests/test_content.py

@@ -29,3 +29,16 @@ def test_isinf(self):
 
     def test_isnan(self):
         self.check_unary_nan("isnan")
+
+
+class TestUfuncLike(unittest.TestCase):
+    @testing.numpy_cupy_array_equal()
+    def check_unary(self, name, xp):
+        a = xp.array([-3, xp.inf, -1, -xp.inf, 0, 1, 2, xp.nan])
+        return getattr(xp, name)(a)
+
+    def test_isneginf(self):
+        self.check_unary("isneginf")
+
+    def test_isposinf(self):
+        self.check_unary("isposinf")