Add new np interfaces

mindspore/numpy/__init__.py

@@ -30,13 +30,14 @@ from .array_ops import (transpose, expand_dims, squeeze, rollaxis, swapaxes, res
                         ravel, concatenate, where, atleast_1d, atleast_2d, atleast_3d,
                         column_stack, hstack, dstack, vstack, stack, unique, moveaxis,
                         tile, broadcast_to, broadcast_arrays, roll, append, split, vsplit,
-                        flip, flipud, fliplr, hsplit, dsplit, take_along_axis, take, repeat)
+                        flip, flipud, fliplr, hsplit, dsplit, take_along_axis, take, repeat,
+                        rot90, select, array_split)
 from .array_creations import copy_ as copy
 from .array_creations import (array, asarray, asfarray, ones, zeros, full, arange,
                               linspace, logspace, eye, identity, empty, empty_like,
                               ones_like, zeros_like, full_like, diagonal, tril, triu,
                               tri, trace, meshgrid, mgrid, ogrid, diagflat,
-                              diag, diag_indices, ix_)
+                              diag, diag_indices, ix_, indices, geomspace, vander)
 from .dtypes import (int_, int8, int16, int32, int64, uint, uint8, uint16,
                      uint32, uint64, float_, float16, float32, float64, bool_, inf, nan,
                      numeric_types, PINF, NINF)
@@ -45,35 +46,51 @@ from .math_ops import (mean, inner, add, subtract, multiply, divide, true_divide
                        matmul, square, sqrt, reciprocal, log, maximum, heaviside, amax, amin,
                        hypot, float_power, floor, ptp, deg2rad, rad2deg, count_nonzero,
                        positive, negative, clip, floor_divide, remainder, fix, fmod, trunc,
-                       exp, expm1, cumsum)
+                       exp, expm1, exp2, kron, promote_types, divmod_, diff, cbrt,
+                       cross, ceil, trapz, gcd, lcm, convolve, log1p, logaddexp, log2,
+                       logaddexp2, log10, ediff1d, nansum, nanmean, nanvar, nanstd, cumsum, nancumsum,
+                       sin, cos, tan, arcsin, arccos, arctan, sinh, cosh, tanh, arcsinh, arccosh,
+                       arctanh, arctan2, cov)
 from .logic_ops import (not_equal, less_equal, less, greater_equal, greater, equal, isfinite,
-                        isnan, isinf, isposinf, isneginf, isscalar)
+                        isnan, isinf, isposinf, isneginf, isscalar, logical_and, logical_not,
+                        logical_or, logical_xor, in1d, isin, isclose)
 
 mod = remainder
 fabs = absolute
+divmod = divmod_ # pylint: disable=redefined-builtin
+abs = absolute # pylint: disable=redefined-builtin
+max = amax # pylint: disable=redefined-builtin
+min = amin # pylint: disable=redefined-builtin
+
 
 array_ops_module = ['transpose', 'expand_dims', 'squeeze', 'rollaxis', 'swapaxes', 'reshape',
                     'ravel', 'concatenate', 'where', 'atleast_1d', 'atleast_2d', 'atleast_3d',
                     'column_stack', 'hstack', 'dstack', 'vstack', 'stack', 'unique', 'moveaxis',
                     'tile', 'broadcast_to', 'broadcast_arrays', 'append', 'roll', 'split', 'vsplit',
                     'flip', 'flipud', 'fliplr', 'hsplit', 'dsplit', 'take_along_axis', 'take',
-                    'repeat']
+                    'repeat', 'rot90', 'select', 'array_split']
 
 array_creations_module = ['array', 'asarray', 'asfarray', 'ones', 'zeros', 'full', 'arange',
                           'linspace', 'logspace', 'eye', 'identity', 'empty', 'empty_like',
                           'ones_like', 'zeros_like', 'full_like', 'diagonal', 'tril', 'triu',
                           'tri', 'trace', 'meshgrid', 'mgrid', 'ogrid', 'diagflat', 'diag',
-                          'diag_indices', 'ix_', 'cumsum']
+                          'diag_indices', 'ix_', 'indices', 'geomspace', 'vander']
 
 math_module = ['mean', 'inner', 'add', 'subtract', 'multiply', 'divide', 'true_divide', 'power',
                'dot', 'outer', 'tensordot', 'absolute', 'std', 'var', 'average', 'not_equal',
                'minimum', 'matmul', 'square', 'sqrt', 'reciprocal', 'log', 'maximum',
                'heaviside', 'amax', 'amin', 'hypot', 'float_power', 'floor', 'ptp', 'deg2rad',
                'rad2deg', 'count_nonzero', 'positive', 'negative', 'clip', 'floor_divide',
-               'remainder', 'mod', 'fix', 'fmod', 'trunc', 'exp', 'expm1', 'fabs', 'cumsum']
+               'remainder', 'mod', 'fix', 'fmod', 'trunc', 'exp', 'expm1', 'fabs', 'exp2', 'kron',
+               'promote_types', 'divmod', 'diff', 'cbrt', 'cross', 'ceil', 'trapz',
+               'abs', 'max', 'min', 'gcd', 'lcm', 'log1p', 'logaddexp', 'log2', 'logaddexp2', 'log10',
+               'convolve', 'ediff1d', 'nansum', 'nanmean', 'nanvar', 'nanstd', 'cumsum',
+               'nancumsum', 'sin', 'cos', 'tan', 'arcsin', 'arccos', 'arctan', 'sinh', 'cosh', 'tanh',
+               'arcsinh', 'arccosh', 'arctanh', 'arctan2', 'cov']
 
 logic_module = ['not_equal', 'less_equal', 'less', 'greater_equal', 'greater', 'equal', 'isfinite',
-                'isnan', 'isinf', 'isposinf', 'isneginf', 'isscalar']
+                'isnan', 'isinf', 'isposinf', 'isneginf', 'isscalar', 'logical_and', 'logical_not',
+                'logical_or', 'logical_xor', 'in1d', 'isin', 'isclose']
 
 __all__ = array_ops_module + array_creations_module + math_module + logic_module + numeric_types
 

mindspore/numpy/dtypes.py

@@ -169,3 +169,16 @@ promotion_rule = {
     (bool_, float32): float32,
     (bool_, float64): float64,
 }
+
+rule_for_trigonometric = {float16: float16,
+                          float32: float32,
+                          float64: float64,
+                          int8: float16,
+                          int16: float32,
+                          int32: float32,
+                          int64: float32,
+                          uint8: float16,
+                          uint16: float32,
+                          uint32: float32,
+                          uint64: float32,
+                          bool_: float16}

mindspore/numpy/utils.py

@@ -13,14 +13,11 @@
 # limitations under the License.
 # ============================================================================
 """internal utility functions"""
-
-import numpy as onp
-
 from ..common import Tensor
 from ..ops import functional as F
 from ..common import dtype as mstype
 
-from .utils_const import _tile_size, _add_unit_axes, _raise_type_error
+from .utils_const import _tile_size, _add_unit_axes, _raise_type_error, _type_convert
 
 
 def _deep_list(array_like):
@@ -56,9 +53,8 @@ def _deep_tensor_to_nparray(array_like):
 
 def _check_input_for_asarray(array_like):
     """check whether array_like argument is a valid type for np.asarray conversion"""
-    if not isinstance(array_like, (Tensor, list, tuple, int, float, bool, onp.ndarray)):
-        _raise_type_error("input data must be `int`, `float`, `bool`, `Tensor`, `list`, `tuple`" + \
-            "or numpy.ndarray, but got ", array_like)
+    if not isinstance(array_like, (Tensor, list, tuple, int, float, bool)):
+        _raise_type_error("input data must be `int`, `float`, `bool`, `Tensor`, `list`, `tuple`, but got ", array_like)
 
 
 def _is_scalar(shape):
@@ -121,6 +117,20 @@ def _convert_64_to_32(tensor):
     return tensor
 
 
+def _to_tensor(*args):
+    """Returns each input as Tensor"""
+    res = ()
+    for arg in args:
+        if isinstance(arg, (int, float, bool, list, tuple)):
+            arg = _convert_64_to_32(_type_convert(Tensor, arg))
+        elif not isinstance(arg, Tensor):
+            _raise_type_error("Expect input to be array like.")
+        res += (arg,)
+    if len(res) == 1:
+        return res[0]
+    return res
+
+
 def _get_dtype_from_scalar(*input_numbers):
     """
     Get the final dtype from series of input numbers, compared with F.typeof, we
@@ -139,3 +149,8 @@ def _get_dtype_from_scalar(*input_numbers):
     if int_flag:
         return mstype.int32
     return mstype.float32
+
+
+def _isnan(x):
+    """Computes isnan."""
+    return F.not_equal(x, x)

mindspore/numpy/utils_const.py

@@ -14,7 +14,8 @@
 # ============================================================================
 """internal graph-compatible utility functions"""
 import math
-from functools import partial
+from itertools import zip_longest
+from collections import deque
 
 import mindspore.context as context
 from ..ops import functional as F
@@ -24,7 +25,7 @@ from ..common import Tensor
 from .._c_expression import Tensor as Tensor_
 from .._c_expression import typing
 
-from .dtypes import promotion_rule, dtype_tuple, all_types, dtype_map
+from .dtypes import promotion_rule, dtype_tuple, all_types, dtype_map, rule_for_trigonometric
 
 
 @constexpr
@@ -110,44 +111,19 @@ def _get_device():
     return context.get_context('device_target')
 
 
-@constexpr
-def _reverse_index(idx, arr):
-    """
-    Returns 1 if shape[idx:] is broadcastable to shape_out[idx:],
-    2 situations if the function returns 1:
-    - 1. Tensor's shape has 1 at the designated dimension.
-    - 2. Tensor's dimension is less than the designated idx. (The Tensor shape
-         has been reversed)
-    For both cases, 2 tensors are broadcastable.
-    otherwise returns the element at position of shape
-    """
-    if len(arr) <= idx:
-        return 1
-    return arr[-1 - idx]
-
-
 @constexpr
 def _infer_out_shape(*shapes):
     """
-    Returns shape of output after broadcasting
-    Raises ValueError if shape1 and shape2 cannot be broadcast
+    Returns shape of output after broadcasting. Raises ValueError if shapes cannot be broadcast.
     """
-    shapes_unbroadcastable = False
-    ndim_max = max(map(len, shapes))
-    shape_out = [0]*ndim_max
-    i = 0
-    for i in range(ndim_max):
-        shape_out[-1 - i] = max(map(partial(_reverse_index, i), shapes))
-        for shape in shapes:
-            if _reverse_index(i, shape) != shape_out[-1 - i]:
-                if _reverse_index(i, shape) != 1:
-                    shapes_unbroadcastable = True
-                    break
-        if shapes_unbroadcastable:
-            break
-    if not shapes_unbroadcastable:
-        return tuple(shape_out)
+    shape_out = deque()
+    reversed_shapes = map(reversed, shapes)
+    for items in zip_longest(*reversed_shapes, fillvalue=1):
+        max_size = 0 if 0 in items else max(items)
+        if any(item not in (1, max_size) for item in items):
             raise ValueError(f'operands could not be broadcast together with shapes {*shapes,}')
+        shape_out.appendleft(max_size)
+    return tuple(shape_out)
 
 
 @constexpr
@@ -228,6 +204,21 @@ def _raise_value_error(info, param=None):
     raise ValueError(info + f"{param}")
 
 
+@constexpr
+def _raise_runtime_error(info, param=None):
+    """
+    Raise RuntimeError in both graph/pynative mode
+
+    Args:
+        info(str): info string to display
+        param(python obj): any object that can be recognized by graph mode. If is
+            not None, then param's value information will be extracted and displayed.
+            Default is None.
+    """
+    if param is None:
+        raise RuntimeError(info)
+    raise RuntimeError(info + f"{param}")
+
 @constexpr
 def _empty(dtype, shape):
     """Returns an uninitialized array with dtype and shape."""
@@ -242,6 +233,9 @@ def _promote(dtype1, dtype2):
         return promotion_rule[dtype1, dtype2]
     return promotion_rule[dtype2, dtype1]
 
+@constexpr
+def _promote_for_trigonometric(dtype):
+    return rule_for_trigonometric[dtype]
 
 @constexpr
 def _max(*args):
@@ -315,7 +309,7 @@ def _canonicalize_axis(axis, ndim):
 
     axis = tuple([canonicalizer(axis) for axis in axis])
     if all(axis.count(el) <= 1 for el in axis):
-        return axis if len(axis) > 1 else axis[0]
+        return tuple(sorted(axis)) if len(axis) > 1 else axis[0]
     raise ValueError(f"duplicate axes in {axis}.")
 
 
@@ -426,13 +420,37 @@ def _tuple_getitem(tup, idx, startswith=True):
 
 
 @constexpr
-def _iota(dtype, num):
+def _tuple_setitem(tup, idx, value):
+    """
+    Returns a tuple with specified `idx` set to `value`.
+    """
+    tup = list(tup)
+    tup[idx] = value
+    return tuple(tup)
+
+
+@constexpr
+def _iota(dtype, num, increasing=True):
     """Creates a 1-D tensor with value: [0,1,...num-1] and dtype."""
     # TODO: Change to P.Linspace when the kernel is implemented on CPU.
+    if increasing:
         return Tensor(list(range(int(num))), dtype)
+    return Tensor(list(range(int(num)-1, -1, -1)), dtype)
 
 
 @constexpr
 def _ceil(number):
     """Ceils the number in graph mode."""
     return math.ceil(number)
+
+
+@constexpr
+def _seq_prod(seq1, seq2):
+    """Returns the element-wise product of seq1 and seq2."""
+    return tuple(map(lambda x, y: x*y, seq1, seq2))
+
+
+@constexpr
+def _make_tensor(val, dtype):
+    """ Returns the tensor with value `val` and dtype `dtype`."""
+    return Tensor(val, dtype)

mindspore/ops/composite/multitype_ops/not_equal_impl.py

@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `not equal` operations."""
 
+from . import _constexpr_utils as const_utils
 from ...composite import base
 from ... import functional as F
 
@@ -41,6 +42,21 @@ def _not_equal_scalar(x, y):
     return not F.scalar_eq(x, y)
 
 
+@not_equal.register("mstype", "mstype")
+def _not_equal_mstype(x, y):
+    """
+    Determine if two mindspore types are not equal.
+
+    Args:
+       x (mstype): first input mindspore type.
+       y (mstype): second input mindspore type.
+
+    Returns:
+       bool, if x != y return true, x == y return false.
+   """
+    return not const_utils.mstype_eq(x, y)
+
+
 @not_equal.register("String", "String")
 def _not_equal_string(x, y):
     """

mindspore/ops/functional.py

@@ -77,6 +77,7 @@ floormod = tensor_mod
 tensor_exp = P.Exp()
 exp = tensor_exp
 tensor_expm1 = P.Expm1()
+tensor_slice = P.Slice()
 strided_slice = P.StridedSlice()
 same_type_shape = P.SameTypeShape()
 check_bprop = P.CheckBprop()
@@ -94,6 +95,22 @@ tensor_slice = P.Slice()
 maximum = P.Maximum()
 minimum = P.Minimum()
 floor = P.Floor()
+logical_not = P.LogicalNot()
+logical_or = P.LogicalOr()
+logical_and = P.LogicalAnd()
+sin = P.Sin()
+cos = P.Cos()
+tan = P.Tan()
+asin = P.Asin()
+acos = P.ACos()
+atan = P.Atan()
+sinh = P.Sinh()
+cosh = P.Cosh()
+tanh = P.Tanh()
+asinh = P.Asinh()
+acosh = P.Acosh()
+atanh = P.Atanh()
+atan2 = P.Atan2()
 
 scalar_to_array = P.ScalarToArray()
 scalar_to_tensor = P.ScalarToTensor()

mindspore/ops/operations/math_ops.py

@@ -2560,7 +2560,7 @@ class Acosh(PrimitiveWithInfer):
         TypeError: If `input_x` is not a Tensor.
 
     Supported Platforms:
-        ``Ascend`` ``GPU`` ``CPU``
+        ``Ascend`` ``GPU``
 
     Examples:
         >>> acosh = ops.Acosh()
@@ -2637,7 +2637,7 @@ class Asinh(PrimitiveWithInfer):
         TypeError: If `input_x` is not a Tensor.
 
     Supported Platforms:
-        ``Ascend`` ``GPU`` ``CPU``
+        ``Ascend`` ``GPU``
 
     Examples:
         >>> asinh = ops.Asinh()

tests/st/numpy_native/test_array_creations.py

@@ -20,7 +20,7 @@ import numpy as onp
 import mindspore.numpy as mnp
 
 from .utils import rand_int, rand_bool, match_array, match_res, match_meta, \
-    match_all_arrays
+    match_all_arrays, run_multi_test, to_tensor
 
 
 class Cases():
@@ -40,8 +40,8 @@ class Cases():
 
         self.array_sets = [1, 1.1, True, [1, 0, True], [1, 1.0, 2], (1,),
                            [(1, 2, 3), (4, 5, 6)], onp.random.random(  # pylint: disable=no-member
-                               (100, 100)).astype(onp.float32),
-                           onp.random.random((100, 100)).astype(onp.bool)]
+                               (100, 100)).astype(onp.float32).tolist(),
+                           onp.random.random((100, 100)).astype(onp.bool).tolist()]
 
         self.arrs = [
             rand_int(2),
@@ -138,8 +138,8 @@ def test_asarray():
             expected = mnp.asarray(array, test_case.mnp_dtypes[i]).asnumpy()
             match_array(actual, expected, error=7)
 
-    # Additional tests for nested tensor/numpy_array mixture
-    mnp_input = [(onp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
+    # Additional tests for nested tensor mixture
+    mnp_input = [(mnp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
     onp_input = [(onp.ones(3,), onp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
 
     actual = onp.asarray(onp_input)
@@ -168,11 +168,11 @@ def test_array():
         assert arr4 is arr5
 
     # Additional tests for nested tensor/numpy_array mixture
-    mnp_input = [(onp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
+    mnp_input = [(mnp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
     onp_input = [(onp.ones(3,), onp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
 
-    actual = onp.asarray(onp_input)
-    expected = mnp.asarray(mnp_input).asnumpy()
+    actual = onp.array(onp_input)
+    expected = mnp.array(mnp_input).asnumpy()
     match_array(actual, expected, error=7)
 
 
@@ -202,11 +202,11 @@ def test_asfarray():
             match_array(actual, expected, error=7)
 
     # Additional tests for nested tensor/numpy_array mixture
-    mnp_input = [(onp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
+    mnp_input = [(mnp.ones(3,), mnp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
     onp_input = [(onp.ones(3,), onp.ones(3)), [[1, 1, 1], (1, 1, 1)]]
 
-    actual = onp.asarray(onp_input)
-    expected = mnp.asarray(mnp_input).asnumpy()
+    actual = onp.asfarray(onp_input)
+    expected = mnp.asfarray(mnp_input).asnumpy()
     match_array(actual, expected, error=7)
 
 
@@ -373,14 +373,14 @@ def test_linspace():
     stop = onp.random.random([1, 5, 1]).astype("float32")
     actual = onp.linspace(start, stop, num=20, retstep=True,
                           endpoint=False, dtype=onp.float32)
-    expected = mnp.linspace(mnp.asarray(start), mnp.asarray(stop), num=20,
+    expected = mnp.linspace(to_tensor(start), to_tensor(stop), num=20,
                             retstep=True, endpoint=False)
     match_array(actual[0], expected[0].asnumpy(), error=6)
     match_array(actual[1], expected[1].asnumpy(), error=6)
 
     actual = onp.linspace(start, stop, num=20, retstep=True,
                           endpoint=False, dtype=onp.int16)
-    expected = mnp.linspace(mnp.asarray(start), mnp.asarray(stop), num=20,
+    expected = mnp.linspace(to_tensor(start), to_tensor(stop), num=20,
                             retstep=True, endpoint=False, dtype=mnp.int16)
     match_array(actual[0], expected[0].asnumpy(), error=6)
     match_array(actual[1], expected[1].asnumpy(), error=6)
@@ -388,7 +388,7 @@ def test_linspace():
     for axis in range(2):
         actual = onp.linspace(start, stop, num=20, retstep=False,
                               endpoint=False, dtype=onp.float32, axis=axis)
-        expected = mnp.linspace(mnp.asarray(start), mnp.asarray(stop), num=20,
+        expected = mnp.linspace(to_tensor(start), to_tensor(stop), num=20,
                                 retstep=False, endpoint=False, dtype=mnp.float32, axis=axis)
         match_array(actual, expected.asnumpy(), error=6)
 
@@ -510,18 +510,18 @@ def test_full_like():
     for mnp_proto, onp_proto in zip(test_case.mnp_prototypes, test_case.onp_prototypes):
         shape = onp.zeros_like(onp_proto).shape
         fill_value = rand_int()
-        actual = mnp.full_like(mnp_proto, mnp.array(fill_value)).asnumpy()
+        actual = mnp.full_like(mnp_proto, to_tensor(fill_value)).asnumpy()
         expected = onp.full_like(onp_proto, fill_value)
         match_array(actual, expected)
 
         for i in range(len(shape) - 1, 0, -1):
             fill_value = rand_int(*shape[i:])
-            actual = mnp.full_like(mnp_proto, mnp.array(fill_value)).asnumpy()
+            actual = mnp.full_like(mnp_proto, to_tensor(fill_value)).asnumpy()
             expected = onp.full_like(onp_proto, fill_value)
             match_array(actual, expected)
 
             fill_value = rand_int(1, *shape[i + 1:])
-            actual = mnp.full_like(mnp_proto, mnp.array(fill_value)).asnumpy()
+            actual = mnp.full_like(mnp_proto, to_tensor(fill_value)).asnumpy()
             expected = onp.full_like(onp_proto, fill_value)
             match_array(actual, expected)
 
@@ -549,6 +549,21 @@ def test_tri_triu_tril():
     match_array(mnp.tri(64, 64, -10).asnumpy(), onp.tri(64, 64, -10))
 
 
+@pytest.mark.level1
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_nancumsum():
+    x = rand_int(2, 3, 4, 5)
+    x[0][2][1][3] = onp.nan
+    x[1][0][2][4] = onp.nan
+    x[1][1][1][1] = onp.nan
+    match_res(mnp.nancumsum, onp.nancumsum, x)
+    match_res(mnp.nancumsum, onp.nancumsum, x, axis=-2)
+    match_res(mnp.nancumsum, onp.nancumsum, x, axis=0)
+    match_res(mnp.nancumsum, onp.nancumsum, x, axis=3)
+
+
 def mnp_diagonal(arr):
     return mnp.diagonal(arr, offset=2, axis1=-1, axis2=0)
 
@@ -653,7 +668,7 @@ def test_meshgrid():
         (2, 3), 9), onp.full((4, 5, 6), 7))
     for i in range(len(xi)):
         arrs = xi[i:]
-        mnp_arrs = map(mnp.asarray, arrs)
+        mnp_arrs = map(to_tensor, arrs)
         for mnp_res, onp_res in zip(mnp_meshgrid(*mnp_arrs), onp_meshgrid(*arrs)):
             match_all_arrays(mnp_res, onp_res)
 
@@ -750,6 +765,68 @@ def test_ix_():
         match_res(mnp_ix_, onp_ix_, *test_arrs)
 
 
+def mnp_indices():
+    a = mnp.indices((2, 3))
+    b = mnp.indices((2, 3, 4), sparse=True)
+    return a, b
+
+
+def onp_indices():
+    a = onp.indices((2, 3))
+    b = onp.indices((2, 3, 4), sparse=True)
+    return a, b
+
+
+def test_indices():
+    run_multi_test(mnp_indices, onp_indices, ())
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_geomspace():
+    start = onp.arange(1, 7).reshape(2, 3)
+    end = [1000, 2000, 3000]
+    match_array(mnp.geomspace(1, 256, num=9).asnumpy(),
+                onp.geomspace(1, 256, num=9), error=1)
+    match_array(mnp.geomspace(1, 256, num=8, endpoint=False).asnumpy(),
+                onp.geomspace(1, 256, num=8, endpoint=False), error=1)
+    match_array(mnp.geomspace(to_tensor(start), end, num=4).asnumpy(),
+                onp.geomspace(start, end, num=4), error=1)
+    match_array(mnp.geomspace(to_tensor(start), end, num=4, endpoint=False).asnumpy(),
+                onp.geomspace(start, end, num=4, endpoint=False), error=1)
+    match_array(mnp.geomspace(to_tensor(start), end, num=4, axis=-1).asnumpy(),
+                onp.geomspace(start, end, num=4, axis=-1), error=1)
+    match_array(mnp.geomspace(to_tensor(start), end, num=4, endpoint=False, axis=-1).asnumpy(),
+                onp.geomspace(start, end, num=4, endpoint=False, axis=-1), error=1)
+
+    start = onp.arange(1, 1 + 2*3*4*5).reshape(2, 3, 4, 5)
+    end = [1000, 2000, 3000, 4000, 5000]
+    for i in range(-5, 5):
+        match_array(mnp.geomspace(to_tensor(start), end, num=4, axis=i).asnumpy(),
+                    onp.geomspace(start, end, num=4, axis=i), error=1)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_vander():
+    arrs = [rand_int(i + 3) for i in range(3)]
+    for i in range(3):
+        mnp_vander = mnp.vander(to_tensor(arrs[i]))
+        onp_vander = onp.vander(arrs[i])
+        match_all_arrays(mnp_vander, onp_vander)
+        mnp_vander = mnp.vander(to_tensor(arrs[i]), N=2, increasing=True)
+        onp_vander = onp.vander(arrs[i], N=2, increasing=True)
+        match_all_arrays(mnp_vander, onp_vander)
+
+
 @pytest.mark.level1
 @pytest.mark.platform_arm_ascend_training
 @pytest.mark.platform_x86_ascend_training

tests/st/numpy_native/test_logic_ops.py

@@ -19,7 +19,8 @@ import numpy as onp
 
 import mindspore.numpy as mnp
 
-from .utils import rand_int, run_binop_test, match_res
+from .utils import rand_int, rand_bool, run_binop_test, run_logical_test, match_res, \
+    match_all_arrays, to_tensor
 
 
 class Cases():
@@ -55,6 +56,15 @@ class Cases():
             rand_int(8, 1, 6, 1)
         ]
 
+        # Boolean arrays
+        self.boolean_arrs = [
+            rand_bool(),
+            rand_bool(5),
+            rand_bool(6, 1),
+            rand_bool(7, 1, 5),
+            rand_bool(8, 1, 6, 1)
+        ]
+
         # array which contains infs and nans
         self.infs = onp.array([[1.0, onp.nan], [onp.inf, onp.NINF], [2.3, -4.5], [onp.nan, 0.0]])
 
@@ -246,10 +256,147 @@ def test_isneginf():
     match_res(mnp_isneginf, onp_isneginf, test_case.infs)
 
 
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
 def test_isscalar():
     assert mnp.isscalar(1) == onp.isscalar(1)
     assert mnp.isscalar(2.3) == onp.isscalar(2.3)
     assert mnp.isscalar([4.5]) == onp.isscalar([4.5])
     assert mnp.isscalar(False) == onp.isscalar(False)
-    assert mnp.isscalar(mnp.array(True)) == onp.isscalar(onp.array(True))
+    assert mnp.isscalar(to_tensor(True)) == onp.isscalar(onp.array(True))
     assert mnp.isscalar('numpy') == onp.isscalar('numpy')
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_isclose():
+    a = [0, 1, 2, float('inf'), float('inf'), float('nan')]
+    b = [0, 1, -2, float('-inf'), float('inf'), float('nan')]
+    match_all_arrays(mnp.isclose(a, b), onp.isclose(a, b))
+    match_all_arrays(mnp.isclose(a, b, equal_nan=True), onp.isclose(a, b, equal_nan=True))
+
+    a = rand_int(2, 3, 4, 5)
+    diff = (onp.random.random((2, 3, 4, 5)).astype("float32") - 0.5) / 1000
+    b = a + diff
+    match_all_arrays(mnp.isclose(to_tensor(a), to_tensor(b), atol=1e-3), onp.isclose(a, b, atol=1e-3))
+    match_all_arrays(mnp.isclose(to_tensor(a), to_tensor(b), atol=1e-3, rtol=1e-4),
+                     onp.isclose(a, b, atol=1e-3, rtol=1e-4))
+    match_all_arrays(mnp.isclose(to_tensor(a), to_tensor(b), atol=1e-2, rtol=1e-6),
+                     onp.isclose(a, b, atol=1e-2, rtol=1e-6))
+
+    a = rand_int(2, 3, 4, 5)
+    b = rand_int(4, 5)
+    match_all_arrays(mnp.isclose(to_tensor(a), to_tensor(b)), onp.isclose(a, b))
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_in1d():
+    xi = [rand_int(), rand_int(1), rand_int(10)]
+    yi = [rand_int(), rand_int(1), rand_int(10)]
+    for x in xi:
+        for y in yi:
+            match_res(mnp.in1d, onp.in1d, x, y)
+            match_res(mnp.in1d, onp.in1d, x, y, invert=True)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_isin():
+    xi = [rand_int(), rand_int(1), rand_int(10), rand_int(2, 3)]
+    yi = [rand_int(), rand_int(1), rand_int(10), rand_int(2, 3)]
+    for x in xi:
+        for y in yi:
+            match_res(mnp.in1d, onp.in1d, x, y)
+            match_res(mnp.in1d, onp.in1d, x, y, invert=True)
+
+
+def mnp_logical_or(x1, x2):
+    return mnp.logical_or(x1, x2)
+
+
+def onp_logical_or(x1, x2):
+    return onp.logical_or(x1, x2)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_logical_or():
+    run_logical_test(mnp_logical_or, onp_logical_or, test_case)
+
+
+def mnp_logical_xor(x1, x2):
+    return mnp.logical_xor(x1, x2)
+
+
+def onp_logical_xor(x1, x2):
+    return onp.logical_xor(x1, x2)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_logical_xor():
+    run_logical_test(mnp_logical_xor, onp_logical_xor, test_case)
+
+
+def mnp_logical_and(x1, x2):
+    return mnp.logical_and(x1, x2)
+
+
+def onp_logical_and(x1, x2):
+    return onp.logical_and(x1, x2)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_logical_and():
+    run_logical_test(mnp_logical_and, onp_logical_and, test_case)
+
+
+def mnp_logical_not(x):
+    return mnp.logical_not(x)
+
+
+def onp_logical_not(x):
+    return onp.logical_not(x)
+
+
+@pytest.mark.level1
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_x86_cpu
+@pytest.mark.env_onecard
+def test_logical_not():
+    for arr in test_case.boolean_arrs:
+        expected = onp_logical_not(arr)
+        actual = mnp_logical_not(to_tensor(arr))
+        onp.testing.assert_equal(actual.asnumpy().tolist(), expected.tolist())

tests/st/numpy_native/utils.py

@@ -15,6 +15,7 @@
 """utility functions for mindspore.numpy st tests"""
 import functools
 import numpy as onp
+from mindspore import Tensor
 import mindspore.numpy as mnp
 
 
@@ -90,7 +91,9 @@ def rand_bool(*shape):
 
 def match_res(mnp_fn, onp_fn, *arrs, **kwargs):
     """Checks results from applying mnp_fn and onp_fn on arrs respectively"""
-    mnp_arrs = map(functools.partial(mnp.asarray, dtype='float32'), arrs)
+    dtype = kwargs.get('dtype', mnp.float32)
+    kwargs.pop('dtype', None)
+    mnp_arrs = map(functools.partial(Tensor, dtype=dtype), arrs)
     error = kwargs.get('error', 0)
     kwargs.pop('error', None)
     mnp_res = mnp_fn(*mnp_arrs, **kwargs)
@@ -151,15 +154,32 @@ def run_unary_test(mnp_fn, onp_fn, test_case, error=0):
 
 
 def run_multi_test(mnp_fn, onp_fn, arrs, error=0):
-    mnp_arrs = map(mnp.asarray, arrs)
+    mnp_arrs = map(Tensor, arrs)
     for actual, expected in zip(mnp_fn(*mnp_arrs), onp_fn(*arrs)):
-        match_array(actual.asnumpy(), expected, error)
+        match_all_arrays(actual, expected, error)
 
 
 def run_single_test(mnp_fn, onp_fn, arr, error=0):
-    mnp_arr = mnp.asarray(arr)
+    mnp_arr = Tensor(arr)
     for actual, expected in zip(mnp_fn(mnp_arr), onp_fn(arr)):
         if isinstance(expected, tuple):
             for actual_arr, expected_arr in zip(actual, expected):
                 match_array(actual_arr.asnumpy(), expected_arr, error)
         match_array(actual.asnumpy(), expected, error)
+
+
+def run_logical_test(mnp_fn, onp_fn, test_case):
+    for x1 in test_case.boolean_arrs:
+        for x2 in test_case.boolean_arrs:
+            match_res(mnp_fn, onp_fn, x1, x2, dtype=mnp.bool_)
+
+def to_tensor(obj, dtype=None):
+    if dtype is None:
+        res = Tensor(obj)
+        if res.dtype == mnp.float64:
+            res = res.astype(mnp.float32)
+        if res.dtype == mnp.int64:
+            res = res.astype(mnp.int32)
+    else:
+        res = Tensor(obj, dtype)
+    return res