torch.logcumsumexp

torch.logcumsumexp(input, dim, *, out=None) → Tensor

Returns the logarithm of the cumulative summation of the exponentiation of elements of input in the dimension dim.

For summation index \(j\) given by dim and other indices \(i\), the result is

\[\text{logcumsumexp}(x)_{ij} = \log \sum\limits_{j=0}^{i} \exp(x_{ij}) \]

Parameters:

• input (Tensor) – the input tensor.

• dim (int) – the dimension to do the operation over

Keyword Arguments:

out (Tensor, optional) – the output tensor.

Example:

>>> a = torch.randn(10)
>>> torch.logcumsumexp(a, dim=0)
tensor([-0.42296738, -0.04462666,  0.86278635,  0.94622083,  1.05277811,
         1.39202815,  1.83525007,  1.84492621,  2.06084887,  2.06844475]))