Expand(), memory keeps increasing

When I run the following code, the memory usage keeps increasing iteration by iteration.

```
while True:
  a = Variable(torch.FloatTensor(32,16).cuda())  
  r = Variable(torch.FloatTensor(1).cuda())
  c = r.expand(a.size())
```

If I use r.exapnd([32, 16]), memory usage does not increase.
Did I use expand() incorrectly? Does anyone have the same issue?

