Merge pull request tensor-compiler#350 from RawnH/pytaco_negation

Adds negation to pytaco tensor interface

python_bindings/pytaco/pytensor/taco_tensor.py

@@ -374,7 +374,10 @@ def __pow__(self, power, modulo=None):
         return tensor_pow(self, power, default_mode)
 
     def __abs__(self):
-        return tensor_abs(self, default_mode)
+        return tensor_abs(self, self.format)
+
+    def __neg__(self):
+        return tensor_neg(self, self.format)
 
     def __array__(self):
         if not _cm.is_dense(self.format):
@@ -1482,6 +1485,39 @@ def tensor_logical_not(t1, out_format, dtype=None):
     """
     return _compute_unary_elt_eise_op(_cm.logical_not, t1, out_format, dtype)
 
+def tensor_neg(t1, out_format, dtype=None):
+    """
+        Negates every value in the tensor.
+
+        The tensor class implements ``__neg__`` using this method.
+
+        Parameters
+        ------------
+        t1: tensor, array_like
+            input tensor or array_like object
+
+        out_format: format, mode_format, optional
+            * If a :class:`format` is specified, the result tensor is stored in the format out_format.
+            * If a :class:`mode_format` is specified, the result the result tensor has a with all of the dimensions
+              stored in the :class:`mode_format` passed in.
+
+        dtype: Datatype
+            The datatype of the output tensor.
+
+
+        Examples
+        ----------
+        >>> import pytaco as pt
+        >>> pt.tensor_neg([1, -2, 0], out_format=pt.dense).toarray()
+        array([-1, 2, 0], dtype=int64)
+
+        Returns
+        --------
+        neg: tensor
+            The element wise negation of the input tensor.
+
+    """
+    return _compute_unary_elt_eise_op(_cm.neg, t1, out_format, dtype)
 
 def tensor_abs(t1, out_format, dtype=None):
     """

python_bindings/unit_tests.py

@@ -251,6 +251,10 @@ def test_mod(self):
         t1[i, j] = pt.remainder(t[i, j], 2)
         self.assertEqual(t1, arr % 2)
 
+    def test_neg(self):
+        arr = np.arange(1, 5).reshape([2, 2])
+        t = pt.from_array(arr)
+        self.assertEqual(-t, -arr)
 
 class testParsers(unittest.TestCase):