Christoph Ortner

An interesting complication is that a `@generated` julia implementation is really extremely fast, but then I can't exploit batched evaluation via AVX or SIMD. So I'm planning to provide two...

OkSo the procedure would be have a Julia package live in a subfolder I think that will then be registered with the Julia General registry. I haven't done this before...

> We have an ongoing offline discussion as to whether we should default to solid harmonics or normalize-by-default I guess for large maxL it won't matter anyhow but for small...

> I feel quite strongly that the splid harmonics are actually the more natural things and it's ok to push for the better option as default I'm coming around to...

I didn't yet test on the same CPU so maybe premature, but I think the Julia codes might be in a similar ballpark as yours. I am guessing I'm somewhere...

Thanks - I'll prepare the PR. I propose to call the subfolder `SpheriCart.jl` which is the [Julia naming convention](https://pkgdocs.julialang.org/v1/creating-packages/#Package-naming-guidelines) since `SpheriCart` will be the name of the package in the...

Hm - maybe one more question. So far I haven't looked at your code at all to avoid license problems. Going forward, if we were to think about optimizing maybe...

> I suggest that for the moment we adapt the Julia API to what we have Initially I'm just going to do SolidHarmonics. I'll add the wrapper for SphericalHarmonics asap....

> Re license I don't mind dual license. Also better to do it now than after we have a bunch of contributors we can't track. My question was more specific...

@Luthaf -- unless you still plan to provide sphericart also via binaries, I think this can now be closed?