Merge pull request #1740 from nouiz/shape_err

theano/tensor/opt.py

@@ -370,12 +370,20 @@ def local_0_dot_x(node):
     if replace:
         constant_zero = T.constant(0, dtype=node.outputs[0].type.dtype)
         if x.ndim == 2 and y.ndim == 2:
+            constant_zero = assert_(constant_zero,
+                                    T.eq(x.shape[1], y.shape[0]))
             return [T.alloc(constant_zero, x.shape[0], y.shape[1])]
         elif x.ndim == 1 and y.ndim == 2:
+            constant_zero = assert_(constant_zero,
+                                    T.eq(x.shape[0], y.shape[0]))
             return [T.alloc(constant_zero, y.shape[1])]
         elif x.ndim == 2 and y.ndim == 1:
+            constant_zero = assert_(constant_zero,
+                                    T.eq(x.shape[1], y.shape[0]))
             return [T.alloc(constant_zero, x.shape[0])]
         elif x.ndim == 1 and y.ndim == 1:
+            constant_zero = assert_(constant_zero,
+                                    T.eq(x.shape[0], y.shape[0]))
             return [constant_zero]
         else:
             _logger.warning("Optimization Warning: "

theano/tensor/tests/test_opt.py

@@ -2377,11 +2377,14 @@ class Test_alloc_zero(unittest.TestCase):
         v2 = tensor.vector('v2')
         m1 = tensor.matrix('m1')
         m2 = tensor.matrix('m2')
-        vv = numpy.asarray([0, 1, 2], dtype=theano.config.floatX)
-        vm = numpy.asarray([[1, 2, 3], [4, 5, 6], [7, 8, 9]],
-                           dtype=theano.config.floatX)
-        for _e1 in [(v1, vv), (m1, vm)]:
-            for _e2 in [(v2, vv), (m2, vm)]:
+        vv2 = numpy.asarray([0, 1], dtype=theano.config.floatX)
+        vm2 = numpy.asarray([[1, 2], [4, 5]],
+                            dtype=theano.config.floatX)
+        vv3 = numpy.asarray([0, 1, 2], dtype=theano.config.floatX)
+        vm3 = numpy.asarray([[1, 2, 3], [4, 5, 6], [7, 8, 9]],
+                            dtype=theano.config.floatX)
+        for _e1 in [(v1, vv2, vv3), (m1, vm2, vm3)]:
+            for _e2 in [(v2, vv2, vv3), (m2, vm2, vm3)]:
                 for p in [0, 1]:
                     if p == 0:
                         e1 = tensor.zeros_like(_e1[0])
@@ -2392,9 +2395,16 @@ class Test_alloc_zero(unittest.TestCase):
                     o = tensor.dot(e1, e2)
                     f = theano.function([_e1[0], _e2[0]], o, mode=self.mode)
                     f(_e1[1], _e2[1])
-                    assert numpy.all([ not isinstance(x.op, tensor.Dot) for x in
+                    f(_e1[2], _e2[2])
+                    assert numpy.all([not isinstance(x.op, tensor.Dot) for x in
                                       f.maker.fgraph.toposort() ])
 
+                    #test that we don't remove shape errors
+                    self.assertRaises((ValueError, AssertionError), f,
+                                      _e1[1], _e2[2])
+                    self.assertRaises((ValueError, AssertionError), f,
+                                      _e1[2], _e2[1])
+
 
 def test_local_subtensor_of_alloc():