Vectorized sum different from looped sum

1

Hi, I am using torch 1.7.1 and I noticed that vectorized sums are different from sums in a loop if the indices are repeated. For example:

import torch

indices = torch.LongTensor([0,1,2,1])
values = torch.FloatTensor([1,1,2,2])
result = torch.FloatTensor([0,0,0])

looped_result = torch.zeros_like(result)

for i in range(indices.shape[0]):
    looped_result[indices[i]] += values[i]

result[indices] += values

print('result',result)
print('looped result', looped_result)

results in:

>> result tensor([1., 2., 2.])
>> looped result tensor([1., 3., 2.])

As you can see the looped variable has the correct sums while the vectorized one doesn’t. Is it possible to avoid the loop and still get the correct result?

2

Since you are using duplicated indices, the behavior is UB and you should use index_put_ instead with accumulate=True:

result.index_put_((indices,),  values, accumulate=True)
3

Thank you very much! it works. I did not know of this method