test lift_transpose_through_dot with broadcastable

theano/tensor/opt.py

@@ -357,7 +357,7 @@ def local_lift_transpose_through_dot(node):
     x, y = node.inputs[0].owner.inputs
 
     if x.ndim == y.ndim == 2:
-        return [broadcast_like(T.dot(y.T, x.T), node.outputs[0], node.env)]
+        return [T.dot(y.T, x.T)]
 
 
 @gof.local_optimizer([])

theano/tensor/tests/test_opt.py

@@ -53,6 +53,27 @@ mode_opt = theano.compile.mode.get_mode(mode_opt)
 ds = lambda x, y: DimShuffle(x.type.broadcastable, y)(x)
 dimshuffle_lift = out2in(local_dimshuffle_lift)
 
+_optimizer_stabilize = gof.Query(include=['fast_run'])
+_optimizer_stabilize.position_cutoff = 1.51
+_optimizer_stabilize = compile.optdb.query(_optimizer_stabilize)
+
+_optimizer_specialize = gof.Query(include=['fast_run'])
+_optimizer_specialize.position_cutoff = 2.01
+_optimizer_specialize = compile.optdb.query(_optimizer_specialize)
+
+_optimizer_fast_run = gof.Query(include=['fast_run'])
+_optimizer_fast_run = compile.optdb.query(_optimizer_fast_run)
+def optimize(g, level='fast_run'):
+    if 'fast_run' is level:
+        _optimizer_fast_run.optimize(g)
+    elif 'specialize' is level:
+        _optimizer_specialize.optimize(g)
+    elif 'stabilize' is level:
+        _optimizer_stabilize.optimize(g)
+    else:
+        raise ValueError(level)
+    return g
+
 
 def inputs(xbc = (0, 0), ybc = (0, 0), zbc = (0, 0)):
     x = TensorType(broadcastable = xbc, dtype = 'float64')('x')
@@ -3089,7 +3110,7 @@ def test_local_div_to_inv():
 
 
 class Test_lift_transpose_through_dot(unittest.TestCase):
-    def optimize(self, g):
+    def simple_optimize(self, g):
         out2in(opt.local_useless_elemwise).optimize(g)
         out2in(opt.local_lift_transpose_through_dot).optimize(g)
         out2in(opt.local_useless_elemwise).optimize(g)
@@ -3097,10 +3118,31 @@ class Test_lift_transpose_through_dot(unittest.TestCase):
 
     def test_matrix_matrix(self):
         a, b = matrices('ab')
-        g = self.optimize(Env([a, b], [tensor.dot(a, b).T]))
+        g = self.simple_optimize(Env([a, b], [tensor.dot(a, b).T]))
         sg = '[dot(InplaceDimShuffle{1,0}(b), InplaceDimShuffle{1,0}(a))]'
         assert str(g) == sg
 
+    def test_row_matrix(self):
+        a = vector('a')
+        b = matrix('b')
+        g = optimize(Env(
+            [a, b],
+            [tensor.dot(a.dimshuffle('x', 0), b).T]),
+            level='stabilize')
+        sg = '[dot(DimShuffle{1,0}(b), DimShuffle{0,x}(a))]'
+        assert str(g) == sg
+
+    def test_matrix_col(self):
+        a = vector('a')
+        b = matrix('b')
+        g = optimize(Env(
+            [a, b],
+            [tensor.dot(b, a.dimshuffle(0, 'x')).T]),
+            level='stabilize')
+        sg = '[dot(DimShuffle{x,0}(a), DimShuffle{1,0}(b))]'
+        assert str(g) == sg
+
+
 if __name__ == '__main__':
 #    unittest.main()
     test_fusion().tes_memory_leak()