MultiCudaTensor++

MultiCudaTensor.lua

@@ -31,6 +31,10 @@ function MCT:size(dim)
    return torch.LongStorage(self._size)
 end
 
+function MCT:dim()
+   return self:size():size()
+end
+
 function MCT:zero()
    for i,tensor in ipairs(self.tensors) do
       cutorch.withDevice(tensor:getDevice(), function()
@@ -277,3 +281,22 @@ function MCT:clone()
    f:close()
    return clone
 end
+
+function MCT:uniform(lower, upper)
+   for i,tensor in ipairs(self.tensors) do
+      cutorch.withDevice(tensor:getDevice(), function() tensor:uniform(lower, upper) end)
+   end
+   return self
+end
+
+-- math.pow(gradParam:norm(),2)
+function MCT:norm(...)
+   assert(#{...} == 0)
+   local norm = 0
+   for i,tensor in ipairs(self.tensors) do
+      norm = norm + cutorch.withDevice(tensor:getDevice(), function() return math.pow(tensor:norm(),2) end)
+   end
+   return math.sqrt(norm)
+end
+
+assert(not MCT.storage, "If you ever define storage, you will need to modify Module.sharedClone in dpnn.Module")

test/test.lua

@@ -156,6 +156,9 @@ function torchxtest.MultiCudaTensor()
    mytester:assert(mweight:size(1) == inputsize)
    mytester:assert(mweight:size(2) == outputsize)
 
+   -- test dim
+   mytester:assert(mweight:dim() == 2)
+
    -- test transpose
    local mwt = mweight:t()
    mytester:assert(mwt.catdim == 1)
@@ -210,6 +213,11 @@ function torchxtest.MultiCudaTensor()
 
    mytester:assertTensorEq(output, output2, 0.0001)
 
+   -- test norm 
+   local norm = mweight:norm()
+   local norm2 = weight:norm()
+   mytester:assert(math.abs(norm - norm2) < 0.0001)
+
    -- test zero
    mweight:zero()
    for i=1,2 do
@@ -244,7 +252,13 @@ function torchxtest.MultiCudaTensor()
    cutorch.withDevice(1, function() mytester:assertTensorEq(mw2.tensors[1], mweight.tensors[1], 0.000001) end)
    cutorch.withDevice(2, function() mytester:assertTensorEq(mw2.tensors[2], mweight.tensors[2], 0.000001) end)
 
+   -- test uniform
+   mw2:uniform(-2, -1)
+   cutorch.withDevice(1, function() mytester:assert(mw2.tensors[1]:min() >= -2 and mw2.tensors[1]:max() <= -1) end)
+   cutorch.withDevice(2, function() mytester:assert(mw2.tensors[2]:min() >= -2 and mw2.tensors[2]:max() <= -1) end)
+
    mytester:assert(cutorch.getDevice() == origdevice)
+
 end
 
 function torchx.test(tests)