Function dfdx::tensor_ops::logsumexp

pub fn logsumexp<T: Tensor<Dtype = f32>>(t: T) -> T::LastDimReduced

Computes the LogSumExp function. Equivalent to log(sum(exp(data))) or data.exp().sum(-1).log().

Calls ln(), sum_last_dim(), and exp()

Examples:

let a = Tensor0D::new(0.0);
let r = logsumexp(a);
assert_eq!(r.data(), &0.0);

let a = Tensor1D::new([-2.0, -1.0, 0.0, 1.0, 2.0]);
let r = a.logsumexp();
assert_eq!(r.data(), &2.4519143);

See log_softmax() and softmax() for related functions.