hongbin

Hi @babbush , I find this [paper](https://arxiv.org/pdf/1111.5337.pdf) list formulas for Fourier transform of Coulomb potential in 1d, 2d, and 3d (on page 2, eq. 2.9), but the result for 2d...

One thing about `@functor` still confuse me. Inside `functor`'s definition, e.g. ```julia functor(x::AbstractArray) = x, y -> y ``` it returns a `Tuple`. The 2nd element of the tuple is...

Hi @willtebbutt , sorry for this late response. I was busy on my project last week and just get time to carefully think about your comment. > The thing I'm...

Cool!, I will collect them into a PR after I finish my assignment this weeks :)

Hi @willtebbutt , the examples works well, thanks ;) > have I covered all of the functionality of interest to GPFlux? Not all, I'm still extending it. The next main...

I found there are related issues [here](https://github.com/willtebbutt/Stheno.jl/issues/51), this can be realized in NKN. Also sparse GP mentioned [here](https://github.com/willtebbutt/Stheno.jl/issues/36) is considered to be added to `GPFlux` in future.

> I think the right way is probably to have a custom NKN kernel that accepts a collection of primitive kernels, and a Chain (or some other Flux construct) that...

Hi @willtebbutt , I just finish some initial work on Gaussian process layer we discussed last week, implementation can be found in this [notebook](https://github.com/HamletWantToCode/GPFlux.jl/blob/develop/notebook/Gaussian_Process_Layer.ipynb) on this [branch](https://github.com/HamletWantToCode/GPFlux.jl/tree/develop). I strongly suggest...

Excellent, it's great to know that `Stheno` has built-in support for this :) Based on our previous discussion, I think we both agree that this Flux integration should include: 1....

> Do you agree / what kinds of resources do you think would be helpful to address this? I think recover some experiments on [deep kernel learning](https://arxiv.org/pdf/1511.02222.pdf) paper is a...