Merge pull request #2548 from carriepl/scan_only_nonseq_inputs

theano/scan_module/scan_op.py

@@ -1347,12 +1347,6 @@ class Scan(PureOp):
             return self.outputs[s:e]
 
         def _get_inner_inps(iidx):
-            s = 0
-            if self.n_seqs > 0:
-                e = 1
-            else:
-                e = len(self.tap_array[0])
-            p = iidx
             if node.inputs[iidx + 1] in self.outer_nitsot(node):
                 return None
             if node.inputs[iidx + 1] in self.outer_non_seqs(node):
@@ -1360,6 +1354,11 @@ class Scan(PureOp):
                     node.inputs[iidx + 1])
                 return [self.inner_non_seqs(self.inputs)[loc_idx]]
 
+            s = 0
+            if self.n_seqs > 0:
+                e = 1
+            else:
+                e = len(self.tap_array[0])
             for p in xrange(iidx):
                 s = e
                 if p < self.n_seqs:

theano/scan_module/tests/test_scan.py

@@ -365,6 +365,27 @@ class T_Scan(unittest.TestCase):
              4,
              numpy.int64([2, 2, 3]))
 
+    @attr('slow')
+    def test_only_nonseq_inputs(self):
+        # Compile the Theano function
+        n_steps=2
+        inp = tensor.matrix()
+        broadcasted_inp, _ = theano.scan(lambda x:x,
+                                         non_sequences=[inp],
+                                         n_steps=n_steps)
+        out = broadcasted_inp.sum()
+        gr = tensor.grad(out, inp)
+        fun = theano.function([inp], [broadcasted_inp, gr])
+
+        # Execute the Theano function and compare outputs to the expected outputs
+        inputs = numpy.array([[1, 2], [3, 4]])
+        expected_out1 = numpy.repeat(inputs[None], n_steps, axis=0)
+        expected_out2 = numpy.ones(inputs.shape, dtype="int8") * n_steps
+
+        out1, out2 = fun(inputs)
+        utt.assert_allclose(out1, expected_out1)
+        utt.assert_allclose(out2, expected_out2)
+
     # simple rnn, one input, one state, weights for each; input/state
     # are vectors, weights are scalars
     def test_one_sequence_one_output_weights(self):