torch.js has not been released yet.

torch.logaddexp

function logaddexp(input: Tensor, other: Tensor, options?: BinaryOptions): Tensor

Computes log(exp(x) + exp(y)) element-wise in a numerically stable way.

\text{logaddexp}(x, y) = \log(\exp(x) + \exp(y))

Numerically stable even for large values

Parameters

Tensor– wise logaddexp

const x = torch.tensor([1.0, 2.0, 3.0]);
const y = torch.tensor([2.0, 3.0, 4.0]);
torch.logaddexp(x, y);  // numerically stable log(exp(x) + exp(y))

function logaddexp(input: Tensor, other: Tensor, options?: BinaryOptions): Tensor

Computes log(exp(x) + exp(y)) element-wise in a numerically stable way.

\text{logaddexp}(x, y) = \log(\exp(x) + \exp(y))

Numerically stable even for large values

Tensor– wise logaddexp

const x = torch.tensor([1.0, 2.0, 3.0]);
const y = torch.tensor([2.0, 3.0, 4.0]);
torch.logaddexp(x, y);  // numerically stable log(exp(x) + exp(y))