This suite turns a Planck noise realization map into a filtered SPIDER map.

For each detector on SPIDER, the steps are:

1. Reobserve with the detector’s beam to obtain a time-ordered product (TOD), a.k.a. a timestream
2. Filter the TOD with a polynomial filter (polyfilter) to get rid off scan-synchronous noise
3. Produce a series of maps for each 3-minute chunkset (“interleaving”)
4. (Optional) Co-add the maps for detectors within the same frequency band

Other than having the latest versions of numpy and spider_tools, this suite makes extensive use of MPI to speed up computations – specifically the mpi4py package.

If you are on a system with a SLURM scheduler, mpi_launcher.sh is your job-submitting script. Modify it as needed. For maximum efficiency, use a loop inside this script to avoid having to manually submit multiple jobs.

It runs through steps 1–3 above in one go by calling mpi_filt_planck_noise_maps.sh. Be sure to check the SBATCH options first!

Once all the detector-specific maps are complete (X1 through X6), run coadd_freq_interleaved_planck_noise_maps.py to perform step 4.

Run cleanup.py to remove files that are no longer needed after the final maps (step 4) are complete. These include the TODs and any intermediary maps produced. While it may be a good idea to keep some TODs around while testing, bear in mind that a single TOD is about 600 MB in size; multiplying by the 1859 “good” channels on SPIDER brings this to about 1 TB.