Skip to content

Latest commit

 

History

History
91 lines (54 loc) · 4.97 KB

File metadata and controls

91 lines (54 loc) · 4.97 KB

High resolution foregrounds in Equatorial with spectral index not varying spatially

Tag: mbs-s0003-20190517

Release date 17 May 2019

This is a re-run of the 201903_highres_foregrounds simulations in Equatorial coordinates instead of Galactic. The only other difference is that hp.smoothing now uses use_pixel_weights=True, i.e. the spherical harmonics transform should be of slightly better quality.

Also, this is the first set of simulations which uses the development version of PySM 3.

The last version of this file is available on Github. The same folder contains all the configuration files used and the scripts to create SLURM jobs.

Input components

Each component is saved separately, all maps are in uK_CMB, IQU or I, single precision (float32)

Dust, synchrotron, free-free and anomalous microwave emission using the "0" models, i.e. SO_d0, SO_s0, SO_f0, and SO_a0, see more details in in the documentation.

These maps are full-sky, you can get the noise maps from the 201901_gaussian_fg_lensed_cmb_realistic_noise simulations to add noise and apply the cut from the scanning strategy. Another option, if you are interested only in the sky cut, is to use the SONoiseSimulator class from mapsims, select nside and scanning_strategy and access the hitmap attribute.

Available maps

HEALPix maps at high resolution (nside 4096) and low resolution (nside 512), these models are deterministic, so we have a set for each resolution for all channels.

Reference frame for the maps is Equatorial. Bandpass for each channel is a delta function centered at the reference frequency. Beams as defined in the SO_Noise_Calculator released in October 2018. The ell_max for the harmonics transform is 3*Nside-1, in the ellmax_2nside subfolder the same simulations were executed just up to 2*Nside.

Location at NERSC:

/project/projectdirs/sobs/v4_sims/mbs/201904_highres_foregrounds_equatorial

The naming convention is:

{output_nside}/{content}/{num:04d}/simonsobs_{content}_uKCMB_{telescope}{band:03d}_nside{nside}_{num:04d}.fits"

where:

  • content is in [cmb, synchtrotron, freefree, ame]
  • num is 0
  • telescope is sa or la
  • band is the channel frequency in GHz

Backed up to tape in ~zonca/sobs/mbs/201904_highres_foregrounds_equatorial.

Available Galactic masks

In Equatorial coordinates

/project/projectdirs/sobs/v4_sims/mbs/201904_highres_foregrounds_equatorial/{output_nside}/masks

Masks are optimized for polarization.

They are generate comparing synchrotron+dust polarized signal at 93 GHz with the CMB standard deviation at one degree angular resolution.

Sky is masked wherever: abs(FG_Q) > thr * std(CMB_Q) or abs(FG_U) > thr * std(CMB_U)

the obtained masks are further smoothed with a gaussian kernel at 10 degree in order to get regular mask edges. The thr parameter ranges from 1 to 4.5. The final masks cover a fraction of the sky (evaluated over the full sky) between 0.28and 0.75

See the Python scripts used to generate the masks in the create_masks folder.

Issues or feedback

In case of any problem with the maps or the documentation or request more/different runs, open an issue on the map_based_simulations repo

Plots

See the plot_maps.ipynb notebook in the folder, executed notebooks:

Software

Performance

At 4096 with lmax to 2*nside all 6 frequency channels: 115 minutes for synchrotron and dust, 95 minutes for freefree and ame, single node on Popeye (SDSC)

At 4096 with lmax to 3*nside-1 each frequency channel (e.g. LA_27, SA_27): 75 minutes for synchrotron and dust, 60 minutes for freefree and ame, single node on Popeye (SDSC)