Add magic method abs to NDArray and Symbol. (#15680)

* add magic method abs to ndarray

* add relevant tests

* add magic method abs to symbol

* add relevant tests

* retrigger CI

* retrigger CI

python/mxnet/ndarray/ndarray.py

@@ -205,6 +205,10 @@ def _to_shared_mem(self):
             self.handle, ctypes.byref(shared_pid), ctypes.byref(shared_id)))
         return shared_pid.value, shared_id.value, self.shape, self.dtype
 
+    def __abs__(self):
+        """x.__abs__() <=> abs(x) <=> x.abs() <=> mx.nd.abs(x, y)"""
+        return self.abs()
+
     def __add__(self, other):
         """x.__add__(y) <=> x+y <=> mx.nd.add(x, y) """
         return add(self, other)

python/mxnet/symbol/symbol.py

@@ -93,6 +93,10 @@ def __iter__(self):
         """
         return (self[i] for i in range(len(self)))
 
+    def __abs__(self):
+        """x.__abs__() <=> abs(x) <=> x.abs() <=> mx.symbol.abs(x, y)"""
+        return self.abs()
+
     def __add__(self, other):
         """x.__add__(y) <=> x+y
 

tests/python/unittest/test_ndarray.py

@@ -172,6 +172,15 @@ def test_ndarray_negate():
     assert_almost_equal(npy, arr.asnumpy())
 
 
+@with_seed()
+def test_ndarray_magic_abs():
+    for dim in range(1, 7):
+        shape = rand_shape_nd(dim)
+        npy = np.random.uniform(-10, 10, shape)
+        arr = mx.nd.array(npy)
+        assert_almost_equal(abs(arr).asnumpy(), arr.abs().asnumpy())
+
+
 @with_seed()
 def test_ndarray_reshape():
     tensor = (mx.nd.arange(30) + 1).reshape(2, 3, 5)

tests/python/unittest/test_symbol.py

@@ -22,7 +22,7 @@
 import numpy as np
 from common import assertRaises, models
 from mxnet.base import NotImplementedForSymbol
-from mxnet.test_utils import discard_stderr
+from mxnet.test_utils import discard_stderr, rand_shape_nd
 import pickle as pkl
 
 def test_symbol_basic():
@@ -188,6 +188,21 @@ def test_symbol_infer_shape_var():
     assert arg_shapes[1] == overwrite_shape
     assert out_shapes[0] == overwrite_shape
 
+
+def test_symbol_magic_abs():
+    for dim in range(1, 7):
+        with mx.name.NameManager():
+            data = mx.symbol.Variable('data')
+            method = data.abs(name='abs0')
+            magic = abs(data)
+            regular = mx.symbol.abs(data, name='abs0')
+            ctx = {'ctx': mx.context.current_context(), 'data': rand_shape_nd(dim)}
+            mx.test_utils.check_consistency(
+                [method, magic], ctx_list=[ctx, ctx])
+            mx.test_utils.check_consistency(
+                [regular, magic], ctx_list=[ctx, ctx])
+
+
 def test_symbol_fluent():
     has_grad = set(['flatten', 'expand_dims', 'flip', 'tile', 'transpose', 'sum', 'nansum', 'prod',
                     'nanprod', 'mean', 'max', 'min', 'reshape', 'broadcast_to', 'split',