I would start with a map with night-only and source-subtracted data at either 90 or 150 GHz (coadded with Planck for the large scales). For example, the relevant map at 90 GHz is "act_planck_dr5.01_s08s18_AA_f090_night_map.fits".
Note that the new version (DR6) will come out very soon, so I imagine any production analysis will be on DR6. Nonetheless, for an early demonstration, we can use DR5 which is public already.
Noise maps:
The noise is quite inhomogeneous, so we have noise maps rather than a single curve. In the link above, there are "ivar" files, that have the inverse variance per pixel. So you can generate a noise realization by drawing independent Gaussian RV for each pixel with mean zero and variance per pixel i is given by sigma^2(i) = 1 / ivar(i) [in muK^2]. DR6 will come with noise simulations with pixel-to-pixel correlations, but for now independent pixels is a good enough assumption.
For example, for the map above, the ivar map is " act_planck_dr5.01_s08s18_AA_f090_night_ivar.fits"
Beams:
Smoothing with a 2D Gaussian beam with FWHM = 1.4 arcmin (90 GHz) or 1.6 (at 150 GHz) is good enough for most things. Also, 1.6 will be exact for the upcoming DR6 data.
Information from Simone on ACT:
Maps: The most recent maps from ACT are here: https://lambda.gsfc.nasa.gov/product/act/actpol_dr5_coadd_maps_get.html
And some more info here: https://lambda.gsfc.nasa.gov/product/act/actpol_dr5_coadd_maps_info.html
I would start with a map with night-only and source-subtracted data at either 90 or 150 GHz (coadded with Planck for the large scales). For example, the relevant map at 90 GHz is "act_planck_dr5.01_s08s18_AA_f090_night_map.fits".
Note that the new version (DR6) will come out very soon, so I imagine any production analysis will be on DR6. Nonetheless, for an early demonstration, we can use DR5 which is public already.
Noise maps: The noise is quite inhomogeneous, so we have noise maps rather than a single curve. In the link above, there are "ivar" files, that have the inverse variance per pixel. So you can generate a noise realization by drawing independent Gaussian RV for each pixel with mean zero and variance per pixel i is given by sigma^2(i) = 1 / ivar(i) [in muK^2]. DR6 will come with noise simulations with pixel-to-pixel correlations, but for now independent pixels is a good enough assumption.
For example, for the map above, the ivar map is " act_planck_dr5.01_s08s18_AA_f090_night_ivar.fits"
Beams: Smoothing with a 2D Gaussian beam with FWHM = 1.4 arcmin (90 GHz) or 1.6 (at 150 GHz) is good enough for most things. Also, 1.6 will be exact for the upcoming DR6 data.
Masks: Appropriate masks are here: https://portal.nersc.gov/project/act/dr6_nilc/ymaps_20230220/masks/
You can start with the footprint-only mask: wide_mask_GAL070_apod_1.50_deg_wExtended.fits
For a real analysis, we'd also at least include the clusters mask: cluster_mask.fits
The other ones are not relevant if we use DR5 data, but will be relevant when we switch to DR6, which should be released in a few months.