• band 9
• less detectors: 252 detectors instead of 684, 7 per wafer, see focalplane plot in #18
• half sampling rate
• multiplied the NET by sqrt(252/684) to get back a noise covariance map and noise map equal to the full focal plane, will scale up the hitmap by 684/252*2~5.4
• using the code in #9

@ziotom78, I've run 2 days with the configuration in #4 for testing purposes.

See /global/cscratch1/sd/zonca/pico/pico_calibration_sim_edison/, this is in global scratch, so you can access it from any machine at NERSC.

I dumped the fits files and quickly checked them, I'll post some plots next.

Generate nside 512 components with tophat bandpasses and symmetric gaussian beams for all 21 frequency bands, using PySM model:

a2,d7,f1,s3

information about bands:
https://zzz.physics.umn.edu/ipsig/baseline

I have a first implementation (suitable for testing only for now) of integrating Conviqt into the satellite pipeline in hpc4cmb/toast#241

I am testing with a simple input map with a signal on the equator:

I have a scanning strategy with no precession, see the hitmap:

For testing I am using a symmetric 10 deg beam prepared by Karl, the results are as expected:

Here is a difference with a map smoothed with healpy:

Get an input FITS file in a predefined format and apply gain to TOAST timelines on the fly

there is a suitable function in toast: https://github.com/hpc4cmb/toast/blob/df23d92042d38f2c280234f0aeb5546272ead93e/src/python/tod/tod_math.py#L19

but it is currently not used in any of the pipelines, need to add support for it

@tskisner @keskitalo

I ran a 100 days simulation of pico but the scanning looks wrong: it always scans the same ring, see the hitmap attached below
Do you know what could be wrong?

My scanning parameters are in https://github.com/zonca/pico-simulations/blob/master/noise_1year/pico_scanning.par

@tskisner @jdborrill, are the new features of the focalplane generation for PICO finalized?
Can you generate the focalplane for the noise sims (or explain me how to generate it)?

• Ran all bands with 7 pixels / wafer and 63 Hz sampling (unless sampling was already below 63 Hz)
• I ran my validation code and checked all the II spectra and they agree with the noise covariance matrix.
• I created the corrected hitmaps

maps are available in:

/global/cscratch1/sd/zonca/pico/full_focal_plane_noise/band_??_sparse_scalednoise_1year/

@jdborrill should I create a folder under /project/projectdirs/pico? what folder name and what naming convention for the files?

2 pixels, 4 detectors at: 0, 45°, 90, e 135°
All boresight pointing.

@ziotom78 still 2 years? With the same configuration of #11?

Number of detectors and scanning parameters changed a bit

https://zzz.physics.umn.edu/ipsig/baseline

Reduce sampling with:

• half sampling rate 126->63
• less detectors per wafer 19->3, so total detectors 684->108
  Reduction is a factor of 12.6666

ran 30 days on 30 nodes

see comparison of hitmaps:
https://gist.github.com/zonca/d9f462502f51c3085604708204a513de

@tskisner @keskitalo, let's take a look at this at the call

For calibration tests, reran the configuration in #11 with only CMB + Solar Dipole + Noise, results available in:

/global/cscratch1/sd/zonca/pico/cal_sims/201804_boresight_1pix_2years_nofg

The spin period for PICO simulation is set to be 60 s, but I believe that TOAST expects this number in minutes (see toast_satellite_sim.py). If this is the case, the following lines of code are wrong:

• https://github.com/zonca/pico-simulations/blob/master/calibration_sim/pico_scanning.par#L4
• https://github.com/zonca/pico-simulations/blob/master/calibration_sim/pico_scanning.par#L8

Apparently the last sample in a file shares the same timestamp as the first sample in the next file:

$ tablist 'tod_fake_0A_science_00000.fits[#row > 10886400]'

                       TIME          THETA            PHI            PSI
   1        86400.0000000000        2.17925        4.62317       -3.08061

             FGTOD         DIPTOD       TOTALTOD  FLAGS GLFLAG
   1    1.31019E-05    0.000178755    8.03019E-05      0      0
$ tablist 'tod_fake_0A_science_00001.fits[#row < 2]'

                       TIME          THETA            PHI            PSI
   1        86400.0000000000        1.55366        4.64231        3.14009

             FGTOD         DIPTOD       TOTALTOD  FLAGS GLFLAG
   1     0.00383942     0.00200384     0.00321391      0      0
$

(strangely, the numbers for PHI, PSI, and so on do not agree).

Unfortunately this makes my code crash:

$ ../dacapo_calibration/index.py index_zonca_185days_0A.ini
[2018-03-01 15:47:44,075 INFO] reading configuration file "index_zonca_185days_0A.ini"
[2018-03-01 15:47:44,077 INFO] configuration file read successfully
[2018-03-01 15:47:44,081 INFO] processing file "tod_fake_0A_science_00000.fits" (1/185)
[2018-03-01 15:48:01,341 INFO] file "tod_fake_0A_science_00000.fits" processed successfully
[2018-03-01 15:48:01,341 INFO] processing file "tod_fake_0A_science_00001.fits" (2/185)
[2018-03-01 15:48:01,377 ERROR] column TIME in HDU 1 of the FITS files is not sorted in ascending order (86400.0 > 86400.00000000001)

Simulation details and output location at NERSC:
https://github.com/zonca/pico-simulations/tree/master/calibration_sim/runs/201804_boresight_1pix_2years

Plot output maps:
https://gist.github.com/zonca/5c86461fd643322d81b2e65c00574361

@ziotom78

Ran on 300 edison nodes using 7.7GB per MPI Process (out of 8 GB), took 13 minutes.

Using configuration in #9

Results in /global/cscratch1/sd/zonca/pico/full_focal_plane_noise/buffered_band_09

Using hpc4cmb/toast#241
I can run some simulations with the real beams provided by Karl.

Please crosscheck the configuration:

• Input sky: we need to prepare a alm file per each detector. If we want to use PySM I can run it beforehand, Galactic foregrounds only, no CMB, no Dipoles, PySM: a2,d7,f1,s3
• Noise: no noise
• Detectors: 2 detectors at 90 deg pointing at boresight
• Band: 150 GHz
• Simulation lenght: 1 simulation of 10 hours, 1 simulation of 1 year