Merge pull request #1113 from huihuifan/narrow_edit

changed narrow to standardize negative length and negative offset beh…

3 files changed, 68 insertions, 59 deletions

diff --git a/Narrow.lua b/Narrow.lua
index a6ebaa3..01be934 100644
--- a/Narrow.lua
+++ b/Narrow.lua
@@ -13,14 +13,16 @@ end
 function Narrow:updateOutput(input)
    local dim = self.dimension < 0 and input:dim() + self.dimension + 1 or self.dimension
    local length = self.length
-   if length < 0 then
-      length = input:size(dim) - self.index + self.length + 2
-   end
    local index = self.index
-   if self.index < 0 then
-      index = 1
-      length = input:size(dim) - length
+
+   if index < 0 then
+      index = input:size(dim) + self.index + 1
    end
+
+   if length < 0 then
+      length = input:size(dim) - index + 1 - torch.abs(length)
+   end
+
    local output=input:narrow(dim, index, length)
    self.output = self.output:typeAs(output)
    self.output:resizeAs(output):copy(output)
@@ -30,14 +32,16 @@ end
 function Narrow:updateGradInput(input, gradOutput)
    local dim = self.dimension < 0 and input:dim() + self.dimension + 1 or self.dimension
    local length = self.length
-   if length < 0 then
-      length = input:size(dim) - self.index + self.length + 2
-   end
    local index = self.index
-   if self.index < 0 then
-      index = 1
-      length = input:size(dim) - length
+
+   if index < 0 then
+      index = input:size(dim) + self.index + 1
    end
+
+   if length < 0 then
+      length = input:size(dim) - index + 1 - torch.abs(length)
+   end
+
    self.gradInput = self.gradInput:typeAs(input)
    self.gradInput:resizeAs(input):zero()
    self.gradInput:narrow(dim,index,length):copy(gradOutput)
diff --git a/doc/simple.md b/doc/simple.md
index 09c60ca..6b0e4d5 100644
--- a/doc/simple.md
+++ b/doc/simple.md
@@ -685,47 +685,45 @@ Narrow is application of [narrow](https://github.com/torch/torch7/blob/master/do
 > x = torch.rand(4, 5)
 
 > x
- 0.3695  0.2017  0.4485  0.4638  0.0513
- 0.9222  0.1877  0.3388  0.6265  0.5659
- 0.8785  0.7394  0.8265  0.9212  0.0129
- 0.2290  0.7971  0.2113  0.1097  0.3166
+ 0.2746  0.8704  0.6839  0.9137  0.5994
+ 0.6099  0.6365  0.0923  0.0795  0.4404
+ 0.3270  0.9202  0.6142  0.8548  0.8239
+ 0.7058  0.6300  0.8553  0.7736  0.3567
 [torch.DoubleTensor of size 4x5]
 
 > nn.Narrow(1, 2, 3):forward(x)
- 0.9222  0.1877  0.3388  0.6265  0.5659
- 0.8785  0.7394  0.8265  0.9212  0.0129
- 0.2290  0.7971  0.2113  0.1097  0.3166
+ 0.6099  0.6365  0.0923  0.0795  0.4404
+ 0.3270  0.9202  0.6142  0.8548  0.8239
+ 0.7058  0.6300  0.8553  0.7736  0.3567
 [torch.DoubleTensor of size 3x5]
 
 > nn.Narrow(1, 2, -1):forward(x)
- 0.9222  0.1877  0.3388  0.6265  0.5659
- 0.8785  0.7394  0.8265  0.9212  0.0129
- 0.2290  0.7971  0.2113  0.1097  0.3166
-[torch.DoubleTensor of size 3x5]
+ 0.6099  0.6365  0.0923  0.0795  0.4404
+ 0.3270  0.9202  0.6142  0.8548  0.8239
+[torch.DoubleTensor of size 2x5]
 
 > nn.Narrow(1, 2, 2):forward(x)
- 0.9222  0.1877  0.3388  0.6265  0.5659
- 0.8785  0.7394  0.8265  0.9212  0.0129
+ 0.6099  0.6365  0.0923  0.0795  0.4404
+ 0.3270  0.9202  0.6142  0.8548  0.8239
 [torch.DoubleTensor of size 2x5]
 
 > nn.Narrow(1, 2, -2):forward(x)
- 0.9222  0.1877  0.3388  0.6265  0.5659
- 0.8785  0.7394  0.8265  0.9212  0.0129
-[torch.DoubleTensor of size 2x5]
-
-> nn.Narrow(2, 2, 3):forward(x)
- 0.2017  0.4485  0.4638
- 0.1877  0.3388  0.6265
- 0.7394  0.8265  0.9212
- 0.7971  0.2113  0.1097
+ 0.6099  0.6365  0.0923  0.0795  0.4404
+[torch.DoubleTensor of size 1x5]
+
+> nn.Narrow(2,2,3):forward(x)
+ 0.8704  0.6839  0.9137
+ 0.6365  0.0923  0.0795
+ 0.9202  0.6142  0.8548
+ 0.6300  0.8553  0.7736
 [torch.DoubleTensor of size 4x3]
 
 > nn.Narrow(2, 2, -2):forward(x)
- 0.2017  0.4485  0.4638
- 0.1877  0.3388  0.6265
- 0.7394  0.8265  0.9212
- 0.7971  0.2113  0.1097
-[torch.DoubleTensor of size 4x3]
+ 0.8704  0.6839
+ 0.6365  0.0923
+ 0.9202  0.6142
+ 0.6300  0.8553
+[torch.DoubleTensor of size 4x2]
 ```
 
 <a name="nn.Replicate"></a>
diff --git a/test.lua b/test.lua
index b19d6b3..f985f2d 100644
--- a/test.lua
+++ b/test.lua
@@ -5854,9 +5854,10 @@ function nntest.Narrow()
    local module1 = nn.Narrow(1, 3, 5)
    local output1 = module1:forward(input)
    local gradInput1 = module1:backward(input, gradOutput)
-   local module2 = nn.Narrow(1, 3, -3)
+   local module2 = nn.Narrow(1, 3, -2)
    local output2 = module2:forward(input)
    local gradInput2 = module2:backward(input, gradOutput)
+
    mytester:assertTensorEq(output, output1, 0.0000001, "Narrow #1 output err")
    mytester:assertTensorEq(gradInput, gradInput1, 0.00001, "Narrow #1 gradInput err")
    mytester:assertTensorEq(output, output2, 0.0000001, "Narrow #1 negative output err")
@@ -5871,7 +5872,7 @@ function nntest.Narrow()
    local module1 = nn.Narrow(2, 5, 3)
    local output1 = module1:forward(input)
    local gradInput1 = module1:backward(input, gradOutput)
-   local module2 = nn.Narrow(2, 5, -4)
+   local module2 = nn.Narrow(2, 5, -3)
    local output2 = module2:forward(input)
    local gradInput2 = module2:backward(input, gradOutput)
    mytester:assertTensorEq(output, output1, 0.0000001, "Narrow #2 output err")
@@ -5888,7 +5889,7 @@ function nntest.Narrow()
    local module1 = nn.Narrow(3, 1, 1)
    local output1 = module1:forward(input)
    local gradInput1 = module1:backward(input, gradOutput)
-   local module2 = nn.Narrow(3, 1, -7)
+   local module2 = nn.Narrow(3, 1, -6)
    local output2 = module2:forward(input)
    local gradInput2 = module2:backward(input, gradOutput)
    mytester:assertTensorEq(output, output1, 0.0000001, "Narrow #3 output err")
@@ -5896,22 +5897,28 @@ function nntest.Narrow()
    mytester:assertTensorEq(output, output2, 0.0000001, "Narrow #3 negative output err")
    mytester:assertTensorEq(gradInput, gradInput2, 0.00001, "Narrow #3 negative gradInput err")
 
-   -- check basic narrow functionality #4
-   local input = torch.rand(3, 10, 4)
-   local output = input:narrow(2, 5, 3)
-   local gradOutput = torch.rand(3, 3, 4)
-   local gradInput = torch.zeros(3, 10, 4)
-   gradInput:narrow(2, 5, 3):copy(gradOutput)
-   local module1 = nn.Narrow(-2, 5, 3)
-   local output1 = module1:forward(input)
-   local gradInput1 = module1:backward(input, gradOutput)
-   local module2 = nn.Narrow(-2, 5, -4)
-   local output2 = module2:forward(input)
-   local gradInput2 = module2:backward(input, gradOutput)
-   mytester:assertTensorEq(output, output1, 0.0000001, "Narrow #4 output err")
-   mytester:assertTensorEq(gradInput, gradInput1, 0.00001, "Narrow #4 gradInput err")
-   mytester:assertTensorEq(output, output2, 0.0000001, "Narrow #4 negative output err")
-   mytester:assertTensorEq(gradInput, gradInput2, 0.00001, "Narrow #4 negative gradInput err")
+   -- check narrow negative offset and negative length
+   local input = torch.Tensor({1, 2, 3, 4, 5})
+
+   local output = torch.Tensor({2})
+   local modelOutput = nn.Narrow(1, 2, 1):forward(input)
+   mytester:assertTensorEq(output, modelOutput, 0.0000001, "Narrow #4.1 output err")
+
+   local output = torch.Tensor({4})
+   local modelOutput = nn.Narrow(1, -2, 1):forward(input)
+   mytester:assertTensorEq(output, modelOutput, 0.0000001, "Narrow #4.2 output err")
+
+   local output = torch.Tensor({5})
+   local modelOutput = nn.Narrow(1, -1, 1):forward(input)
+   mytester:assertTensorEq(output, modelOutput, 0.0000001, "Narrow #4.3 output err")
+
+   local output = torch.Tensor({4})
+   local modelOutput = nn.Narrow(1, -2, -1):forward(input)
+   mytester:assertTensorEq(output, modelOutput, 0.0000001, "Narrow #4.4 output err")
+
+   local output = torch.Tensor({2, 3, 4})
+   local modelOutput = nn.Narrow(1, 2, -1):forward(input)
+   mytester:assertTensorEq(output, modelOutput, 0.0000001, "Narrow #4.5 output err")
 
    -- check narrow negative offset
    local input = torch.rand(3, 10, 4)
@@ -5919,7 +5926,7 @@ function nntest.Narrow()
    local gradOutput = torch.rand(3, 3, 4)
    local gradInput = torch.zeros(3, 10, 4)
    gradInput:narrow(2, 1, 3):copy(gradOutput)
-   local module1 = nn.Narrow(2, -1, 7)
+   local module1 = nn.Narrow(2, -10, 3)
    local output1 = module1:forward(input)
    local gradInput1 = module1:backward(input, gradOutput)
    local module2 = nn.Narrow(2, 1, 3)