[API] Add positive

python/mxnet/ndarray/numpy/_op.py

@@ -50,7 +50,8 @@
            'true_divide', 'nonzero', 'quantile', 'percentile', 'shares_memory', 'may_share_memory', 'interp',
            'diff', 'ediff1d', 'resize', 'polyval', 'nan_to_num', 'isnan', 'isinf', 'isposinf', 'isneginf', 'isfinite',
            'atleast_1d', 'atleast_2d', 'atleast_3d', 'fill_diagonal', 'squeeze',
-           'where', 'bincount', 'rollaxis', 'diagflat', 'repeat', 'prod', 'pad', 'cumsum', 'sum', 'diag', 'diagonal']
+           'where', 'bincount', 'rollaxis', 'diagflat', 'repeat', 'prod', 'pad', 'cumsum', 'sum', 'diag', 'diagonal',
+           'positive']
 
 
 @set_module('mxnet.ndarray.numpy')
@@ -3561,6 +3562,40 @@ def negative(x, out=None, **kwargs):
     return _pure_unary_func_helper(x, _api_internal.negative, _np.negative, out=out)
 
 
+@set_module('mxnet.ndarray.numpy')
+@wrap_np_unary_func
+def positive(x, out=None, **kwargs):
+    r"""
+    Computes the numerical positive of each element `x_i` (i.e.,`y_i = +x_i`)
+    of the input array x .
+
+    Parameters
+    ----------
+    x : ndarray or scalar
+        Input array.
+
+    Returns
+    -------
+    y : ndarray or scalar
+        Returned array or scalar: y = +x. This is a scalar if x is a scalar.
+
+    Notes
+    -----
+    Equivalent to `x.copy()`, but only defined for types that support arithmetic.
+
+    Examples
+    --------
+    >>> x1 = np.array(([1., -1.]))
+    >>> np.positive(x1)
+    array([ 1., -1.])
+    >>> +x1
+    array([ 1., -1.])
+    """
+    if out is x:
+        return x
+    return _pure_unary_func_helper(x, _api_internal.copy, _np.positive, out=out)
+
+
 @set_module('mxnet.ndarray.numpy')
 @wrap_np_unary_func
 def fix(x, out=None, **kwargs):

python/mxnet/numpy/multiarray.py

@@ -80,7 +80,7 @@
            'quantile', 'percentile', 'shares_memory', 'may_share_memory', 'diff', 'ediff1d', 'resize', 'matmul',
            'nan_to_num', 'isnan', 'isinf', 'isposinf', 'isneginf', 'isfinite', 'polyval', 'where', 'bincount',
            'atleast_1d', 'atleast_2d', 'atleast_3d', 'fill_diagonal', 'squeeze',
-           'diagflat', 'repeat', 'prod', 'pad', 'cumsum', 'sum', 'rollaxis', 'diag', 'diagonal']
+           'diagflat', 'repeat', 'prod', 'pad', 'cumsum', 'sum', 'rollaxis', 'diag', 'diagonal', 'positive']
 
 __all__ += fallback.__all__
 
@@ -1094,8 +1094,13 @@ def __mul__(self, other):
         return multiply(self, other)
 
     def __neg__(self):
+        """x.__neg__() <=> -x"""
         return negative(self)
 
+    def __pos__(self):
+        """x.__pos__() <=> +x"""
+        return positive(self)
+
     @wrap_mxnp_np_ufunc
     def __imul__(self, other):
         r"""x.__imul__(y) <=> x \*= y"""
@@ -4823,6 +4828,38 @@ def negative(x, out=None, **kwargs):
     return _mx_nd_np.negative(x, out=out)
 
 
+@set_module('mxnet.numpy')
+@wrap_np_unary_func
+def positive(x, out=None, **kwargs):
+    r"""
+    Computes the numerical positive of each element `x_i` (i.e.,`y_i = +x_i`)
+    of the input array x .
+
+    Parameters
+    ----------
+    x : ndarray or scalar
+        Input array.
+
+    Returns
+    -------
+    y : ndarray or scalar
+        Returned array or scalar: y = +x. This is a scalar if x is a scalar.
+
+    Notes
+    -----
+    Equivalent to `x.copy()`, but only defined for types that support arithmetic.
+
+    Examples
+    --------
+    >>> x1 = np.array(([1., -1.]))
+    >>> np.positive(x1)
+    array([ 1., -1.])
+    >>> +x1
+    array([ 1., -1.])
+    """
+    return _mx_nd_np.positive(x, out=out)
+
+
 @set_module('mxnet.numpy')
 @wrap_np_unary_func
 def fix(x, out=None, **kwargs):

tests/python/unittest/test_numpy_op.py

@@ -2818,6 +2818,7 @@ def forward(self, a, *args, **kwargs):
         'absolute' : (lambda x: -1. * (x < 0) + (x > 0), -1.0, 1.0),
         'logical_not' : (None, -1.0, 1.0),
         'negative' : (lambda x: -1. * onp.ones(x.shape), -1.0, 1.0),
+        'positive' : (lambda x: onp.ones(x.shape), -1.0, 1.0),
         'reciprocal' : (lambda x: -1. / (x ** 2), 0.01, 1.0),
         'sign' : (None, -1.0, 1.0),
         'square' : (lambda x: 2.0 * x, -1.0, 1.0),