Make squared l2 distance?

[vnmabus]

Motivation

In some cases, such as in the Gaussian (RBF) kernel we compute the squared of the L2 distance, which undoes the last sqrt made in the distance.

Desired functionality

We should analyze if we want to provide a squared l2 distance, similar to sqeuclidean in Scipy.

Alternatives

No response

Additional context

No response

[eliegoudout]

I know it's not really conventional, but imo, it'd be simpler to add an optional kw raised: bool = False to functions like norm, disabling the final ** 1 / p op. Unless some optimization magic makes it so that the function doesn't actually use ** 1 / p at the end.