Bug or feature for longTensor?

1 
import torch
a = (torch.rand(3,4)*10).long()
print(a)
print(a*0.9)

Output:

tensor([[8, 5, 3, 4],
        [1, 3, 6, 7],
        [8, 5, 8, 8]])
tensor([[0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0]])
2

Hi,

I don’t think thats a bug, it tries to retain the type of the tensor as long hence the rounding of 0.9 to 0.