Comparison of pDB and DSR noise levels

[zonca]

The new simulations 202102_design_tool_run have different noise levels compared to the DSR. See the plots at https://github.com/CMB-S4/s4mapbasedsims/blob/master/202102_design_tool_run/plots/SAT.md#polarization

pDB / DSR N_ell in simulations

this is polarization N_ell ratio

LF: 1.14, 0.99
MFL: 1.43 / 0.91
MFH: 1.34 / 0.89
HF: 0.44 / 1.21

pDB / DSR N_ell expected from NET and number of detectors

LF: 0.89 / 0.81
MF1: 1.16 / 0.83
MF2: 1.10 / 0.81
HF: 0.66 / 1.25

Take FOV into account

if we also include the effect of a reduction in FOV, for HF FOV=40.74 to FOV=29.32

Very roughly speaking, the area of the observed patch scales like FOV^2. Given a fixed observing time, N_ell is proportional to the sky area. Combining the two, N_ell is proportional to FOV^2.

So this estimate gives (29.32/40.74)**2 = 0.52

[zonca]

@keskitalo I have updated the comparison here with the new simulation run. We need to be able to explain the differences.

[zonca]

@keskitalo @smsimon

Ok I have computed all the ratios, they seem reasonable:

|   band | NET_ratio | num_detectors_ratio | efficiency_ratio | FOV_ratio |
| ULFPL1 |     0.702 |               1.000 |            1.743 |     1.000 |
|   LFL1 |     0.641 |               1.000 |            0.999 |     1.000 |
|  LFPL1 |     0.626 |               1.000 |            1.743 |     1.000 |
|   LFL2 |     0.824 |               1.000 |            0.999 |     1.000 |
|  LFPL2 |     0.839 |               1.000 |            1.743 |     1.000 |
|  MFPL1 |     0.822 |               1.000 |            1.743 |     1.000 |
|   MFL1 |     0.823 |               1.000 |            0.999 |     1.000 |
|   MFL2 |     0.727 |               1.000 |            0.999 |     1.000 |
|  MFPL2 |     0.707 |               1.000 |            1.625 |     1.000 |
|   HFL1 |     0.789 |               1.000 |            0.881 |     1.000 |
|  HFPL1 |     0.806 |               1.000 |            1.602 |     1.000 |
|   HFL2 |     0.768 |               1.000 |            0.881 |     1.000 |
|  HFPL2 |     0.763 |               1.000 |            1.107 |     1.000 |
|   LFS1 |     0.956 |               0.973 |            0.779 |     1.000 |
|   LFS2 |     0.912 |               0.973 |            0.779 |     1.000 |
|  MFLS1 |     1.074 |               0.992 |            0.779 |     1.000 |
|  MFHS1 |     1.043 |               0.992 |            0.779 |     1.000 |
|  MFLS2 |     0.907 |               0.992 |            0.622 |     1.000 |
|  MFHS2 |     0.898 |               0.992 |            0.622 |     1.000 |
|   HFS1 |     0.815 |               1.000 |            0.442 |     0.720 |
|   HFS2 |     1.120 |               1.000 |            0.306 |     0.720 |

Next I'll use this to compute the N_ell ratio as:

 N_ell_pdb / N_ell_dsr = (NET_pbd / NET_dsr * Eff_dsr / Eff_pdb * FOV_pdb/FOV_dsr) ** 2 * (ndet_dsr / ndet_pbd) 

and correct the Reference design sensitivity values by this factor.

[zonca]

I switched to comparing with PBDR, for SAT see:

https://github.com/CMB-S4/s4mapbasedsims/blob/master/202102_design_tool_run/plots/SAT.md

@keskitalo @smsimon there are still significant differences, do I need to apply any corrections to the PBDR noise levels to be comparable with the simulations?

[zonca]

need to compare with http://bicep.rc.fas.harvard.edu/CMB-S4/analysis_logbook/20210506_dt1_vs_bk15_3/

by eye the values in figure 3 of the plot agree with the values I extracted from PBDR added to the instrument model: https://github.com/CMB-S4/s4mapbasedsims/blob/master/202102_design_tool_run/instrument_model/cmbs4_instrument_model.tbl

[zonca]

Ok, I was able to reproduce Clem's plots, they agree:

I will add these plots to the documentation.