Luca Palmieri

Thanks for the review! TODOs are gone, the generic issue has been solved and the panics have been polished. It should be ready for a second look 😁

I am actually happy to work on the shim myself, to get familiar with the crate. I'll spend some time on it over the Christmas break and I'll pull you...

> I am currently working on adding `ndarray`'s `ArrayBase` as an optional input type which should give you all the building blocks needed (currently fighting with the `Data` trait...). >...

Ok, I have started to write the shim, which means I looked closer at the whole crate and I am struggling to understand some design choices. Let's start from my...

It's there any reference (book/article) I can look at for the training algorithm you are using?

Ok, let me rephrase then. You are saying that most projects work under the assumption that `m(x)` in  is `0`, simplifying the GP definition to  while `friedrich` lets...

Perfect. I have started to draft the shim - you can follow my progresses here: https://github.com/LukeMathWalker/linfa/blob/gp/linfa-gaussian-processes/src/hyperparameters.rs

> I see that you put a `???` next to `convergence_fraction` (I should probably find a better name for this parameter). > It corresponds to the current stopping criteria for...

Yeah, when I say `make it work` I mean from an API point of view :+1:

I end up building something like this in almost every project I use `actix-web` with ❤️