@lpsinger @cdeil - I did an update of these methods for the nested and ring scheme. Now it handles numpy broadcasting feature. I do the broadcasting in python side (by using the numpy broadcast_arrays method https://docs.scipy.org/doc/numpy/reference/generated/numpy.broadcast_arrays.html) as broadcasting is not fully implemented in ndarray crate (as seen here). I then send the broadcasted arrays to the rust extension code that will process them as normal array views.
I do have a 3x speedup factor between cdshealpix.healpix_to_lonlat and astropy_healpix.healpix_to_lonlat for 1000 HEALPix indices broadcasted over 10 different depths.
This refers to astropy/astropy-healpix#128
@lpsinger @cdeil - I did an update of these methods for the nested and ring scheme. Now it handles numpy broadcasting feature. I do the broadcasting in python side (by using the numpy broadcast_arrays method https://docs.scipy.org/doc/numpy/reference/generated/numpy.broadcast_arrays.html) as broadcasting is not fully implemented in ndarray crate (as seen here). I then send the broadcasted arrays to the rust extension code that will process them as normal array views.
I do have a 3x speedup factor between cdshealpix.healpix_to_lonlat and astropy_healpix.healpix_to_lonlat for 1000 HEALPix indices broadcasted over 10 different depths.