Using healpy spin transforms in a quicker way

Question

I'm using a code that needs to do frequent spin-2 transforms in healpy using both map2alm and alm2map (predominantly the latter). However, when I go to higher and higher nside this transform inevitably becomes slower.

I am using healpy with openmp and due to the sheer volume of transforms necessary, this still results in a long time to complete a job. Is there any further way to try and speed up the transform that might be applicable to this situation? For example, as I am only interested in part of the sky, is there a way to only reconstruct the values in specified pixels and not the whole sky when using alm2map, and would this be quicker?

Thanks for any help

Answer 1

I don't think healpy can sped this up, but you might be interested in checking out the other Cl estimators, in case you're interested in the power spectra and not only the alm.

For the full sky (in HEALPix):

• PolSpice
• XPol

For smaller fields:

• Poker

For reasons described here, the estimation of the power spectra is not straightforward for a partially masked sky. HEALPy does not correct for this; the other packages I linked above do.