Merge pull request #1701 from nouiz/local_greedy_distributor

theano/tensor/nnet/conv3d2d.py

@@ -164,6 +164,8 @@ def conv3d(signals, filters,
            border_mode='valid'):
     """Convolve spatio-temporal filters with a movie.
 
+    It flip the filters.
+
     :param signals: timeseries of images whose pixels have color channels.
             shape: [Ns, Ts, C, Hs, Ws]
     :param filters: spatio-temporal filters
@@ -173,6 +175,9 @@ def conv3d(signals, filters,
     :param border_mode: The only one tested is 'valid'.
 
     :note: Work on the GPU.
+           Another way to define signals: (batch,  time, in channel, row, column)
+           Another way to define filters: (out channel,time,in channel, row, column)
+
     """
 
     if isinstance(border_mode, str):

theano/tensor/opt.py

@@ -3936,7 +3936,7 @@ def attempt_distribution(factor, num, denum):
                                    neg_pairs))), num, denum
 
 
-@gof.local_optimizer([T.mul, T.true_div])
+@gof.local_optimizer([T.mul, T.true_div, T.inv])
 def local_greedy_distributor(node):
     """
     This optimization tries to apply distributivity of multiplication

theano/tensor/tests/test_blas.py

@@ -19,9 +19,8 @@ from theano.tensor.blas import (_dot22, _dot22scalar, res_is_a, _as_scalar,
                                 gemm_inplace, gemm_no_inplace,
                                 InconsistencyError, Ger, ger, ger_destructive)
 from theano.tests import unittest_tools
-from test_basic import (_approx_eq, as_tensor_variable, inplace_func,
+from test_basic import (as_tensor_variable, inplace_func,
                         compile, inplace)
-        #, constant, eval_outputs)
 import theano.tensor.blas_scipy
 
 
@@ -50,7 +49,6 @@ class t_gemm(TestCase):
     """
     def setUp(self):
         unittest_tools.seed_rng()
-        _approx_eq.debug = 0
         Gemm.debug = False
 
     @staticmethod
@@ -84,8 +82,7 @@ class t_gemm(TestCase):
                 z_after = self._gemm(z_orig, a, x, y, b)
 
                 #print z_orig, z_after, z, type(z_orig), type(z_after), type(z)
-                #_approx_eq.debug = 1
-                self.assertTrue(_approx_eq(z_after, z))
+                unittest_tools.assert_allclose(z_after, z)
                 if a == 0.0 and b == 1.0:
                     return
                 elif z_orig.size == 0:
@@ -150,7 +147,6 @@ class t_gemm(TestCase):
                  self.rand(3, 5), self.rand(5, 4), -1.0)
 
     def test10(self):
-        _approx_eq.debug = 1
         self.cmp(self.rand(3, 4), -1.0, self.rand(3, 5), self.rand(5, 4), 0.0)
 
     def test11(self):
@@ -281,14 +277,11 @@ class t_gemm(TestCase):
             f = inplace_func([], gemm_inplace(tz, ta, tx, ty, tb),
                              mode=compile.Mode(optimizer=None, linker=l))
             f()
-            self.assertTrue(_approx_eq(z_after, tz.get_value(borrow=True)),
-                            (z_orig, z_after, z, z_after - z))
+            unittest_tools.assert_allclose(z_after, tz.get_value(borrow=True))
             f()
-            self.assertTrue(_approx_eq(z_after, tz.get_value(borrow=True)),
-                            (z_orig, z_after, z, z_after - z))
+            unittest_tools.assert_allclose(z_after, tz.get_value(borrow=True))
             f()
-            self.assertTrue(_approx_eq(z_after, tz.get_value(borrow=True)),
-                            (z_orig, z_after, z, z_after - z))
+            unittest_tools.assert_allclose(z_after, tz.get_value(borrow=True))
 
             #tz.value *= 0 # clear z's value
             y_T = ty.get_value(borrow=True).T
@@ -298,7 +291,7 @@ class t_gemm(TestCase):
             f()
             # test that the transposed version of multiplication gives
             # same answer
-            self.assertTrue(_approx_eq(z_after, tz.get_value(borrow=True).T))
+            unittest_tools.assert_allclose(z_after, tz.get_value(borrow=True).T)
 
         t(C, A, B)
         t(C.T, A, B)