Split (join grad), now support split of 0 element.

theano/tensor/basic.py

@@ -3150,8 +3150,9 @@ class Split(Op):
         if numpy.sum(splits) != len_along_axis:
             raise ValueError('The splits sum to %s, expected %s' %
                              (numpy.sum(splits), len_along_axis))
-        if not python_all(splits):
-            raise ValueError('Cannot have a split of zero.')
+        if python_any([nb < 0 for nb in splits]):
+            raise ValueError('Split: you try to make an ndarray with'
+                             'negative number of elements.')
 
         # Checking is done, let's roll the splitting algorithm!
         # Basically we step along the given axis of x, extracting

theano/tensor/tests/test_basic.py

@@ -3547,6 +3547,22 @@ class T_Join_and_Split(unittest.TestCase):
         m = self.shared(rng.rand(4, 4).astype(self.floatX))
         self.assertRaises(TypeError, self.join_op(), 0, v, m)
 
+    def test_split_0elem(self):
+        rng = numpy.random.RandomState(seed=utt.fetch_seed())
+        m = self.shared(rng.rand(4, 6).astype(self.floatX))
+        o = self.split_op(2)(m, 0, [4, 0])
+        f = function([], o, mode=self.mode)
+        o1, o2 = f()
+        assert numpy.allclose(o1, m.get_value(borrow=True))
+        assert numpy.allclose(o2, m.get_value(borrow=True)[4:])
+
+    def test_split_neg(self):
+        rng = numpy.random.RandomState(seed=utt.fetch_seed())
+        m = self.shared(rng.rand(4, 6).astype(self.floatX))
+        o = self.split_op(2)(m, 0, [5, -1])
+        f = function([], o, mode=self.mode)
+        self.assertRaises(ValueError, f)
+
 
 class test_comparison(unittest.TestCase):
     """Test <, >, <=, >=, == and !=