Matt Price

Hi @mreineck, as far as I can tell it seems as though you're calling everything correctly. @jasonmcewen and I have noticed similar behaviour and it's something we plan to look...

Fantastic, thanks @mreineck I'll try and find some time soon to look into this.

@matt-graham I had some issues here relating to healpy dropping support for older MacOS versions. I also ran into a version control issue relating to SciPy and healpy, a solution...

> oookkkk I admit defeat.... I can't figure out how to do it, I was hoping some clever uses of https://jax.readthedocs.io/en/latest/_autosummary/jax.lax.dynamic_slice.html would work, but no, the size has to be...

Hey @lgrcia so this is something we've been wondering about internally, mainly from an implementation point of view (i.e. whether to wrap the JAX backend or just translate it outright)....

Hi @EiffL the first error rises because we expect `L >= 2*nside`. We had a assert for this originally but this must have been lost during development, so definitely something...

If each ring had the same number of samples then the HEALPix FFT should just reduce to a simple `jnp.fft ` call which should solve your compilation time issue. What...

Hi @lgrcia, sorry I couldn't look at this sooner. Unfortunately, I can't see any obvious reason why the three term `einsum` in `impl` should be slower than the full recursion....