more corrections, grad not work still

theano/tensor/signal/downsample.py

@@ -223,7 +223,7 @@ class DownsampleFactorMax(Op):
         # pad the image
         fill = x.min()-1
         y = numpy.zeros((x.shape[0], x.shape[1], img_rows, img_cols)) + fill
-        y[:, :, pad_h:(y.shape[2]-pad_h), pad_w:(y.shape[3]-pad_w)] = x
+        y[:, :, pad_h:(img_rows-pad_h), pad_w:(img_cols-pad_w)] = x
         # max pooling
         for n in xrange(x.shape[0]):
             for k in xrange(x.shape[1]):
@@ -374,7 +374,7 @@ class DownsampleFactorMaxGrad(Op):
         # pad the image
         fill = x.min()-1
         y = numpy.zeros((x.shape[0], x.shape[1], img_rows, img_cols)) + fill
-        y[:, :, pad_h:(y.shape[2]-pad_h), pad_w:(y.shape[3]-pad_w)] = x
+        y[:, :, pad_h:(img_rows-pad_h), pad_w:(img_cols-pad_w)] = x
         gx = numpy.zeros_like(y)
         for n in xrange(x.shape[0]):
             for k in xrange(x.shape[1]):
@@ -389,7 +389,7 @@ class DownsampleFactorMaxGrad(Op):
                                 if (maxout[n, k, r, c] == y[n, k, row_ind, col_ind]):
                                     gx[n, k, row_ind, col_ind] += gz[n, k, r, c]
         # unpad the image
-        gx = gx[:, :, pad_h:(y.shape[2]-pad_h), pad_w:(y.shape[3]-pad_w)]
+        gx = gx[:, :, pad_h:(img_rows-pad_h), pad_w:(img_cols-pad_w)]
         gx_stg[0] = gx
 
     def infer_shape(self, node, in_shapes):

theano/tensor/signal/tests/test_downsample.py

@@ -265,7 +265,7 @@ class TestDownsampleFactorMax(utt.InferShapeTester):
             
     def test_DownsampleFactorMaxPaddingStride_grad(self):
         rng = numpy.random.RandomState(utt.fetch_seed())
-        imval = rng.rand(10, 10, 10, 10) * 10.0
+        imval = rng.rand(1, 1, 10, 10) * 10.0
         maxpoolsize = (5, 3)
         stridesize = (3, 2)
         paddingsize = (2,2)