fixed randomstreams module to be shared when it occurs multiple times in a module graph

theano/tensor/randomstreams.py

@@ -109,7 +109,9 @@ class RandomStreams(Component):
     def allocate(self, memo):
         """override `Component.allocate` """
         for old_r, new_r in self.random_state_variables:
-            assert old_r not in memo
+            if old_r in memo:
+                assert memo[old_r].update is new_r
+            else:
                 memo[old_r] = In(old_r, 
                         value=Container(old_r, storage=[None]),
                         update=new_r,
@@ -117,7 +119,10 @@ class RandomStreams(Component):
 
     def build(self, mode, memo):
         """override `Component.build` """
-        return RandomStreamsInstance(self, memo, self.default_instance_seed)
+        if self not in memo:
+            print 'creating RandomStreamsInstance'
+            memo[self] = RandomStreamsInstance(self, memo, self.default_instance_seed)
+        return memo[self]
 
     def gen(self, op, *args, **kwargs):
         """Create a new random stream in this container.

theano/tensor/tests/test_randomstreams.py

@@ -122,6 +122,22 @@ class T_RandomStreams(unittest.TestCase):
         assert numpy.all(fn_val0 == numpy_val0)
         assert numpy.all(fn_val1 == numpy_val1)
 
+    def test_multiple(self):
+        M = Module()
+        M.random = RandomStreams(234)
+        out = M.random.uniform((2,2))
+        M.m2 = Module()
+        M.m2.random = M.random
+        out2 = M.m2.random.uniform((2,2))
+        M.fn = Method([], out)
+        M.m2.fn2 = Method([], out2)
+        m = M.make()
+        m.random.initialize()
+        m.m2.initialize()
+
+        assert m.random is m.m2.random
+
+
 
 if __name__ == '__main__':
     from theano.tests import main