Difference between revisions of "N eff Forecasting Comparison with DRAFT ILC"

From CMB-S4 wiki
Jump to navigationJump to search
 
(25 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
===Overview===
 
===Overview===
 +
This posting summarises the DRAFT (Dark Radiation Anisotropy Flowdown Team) tool and the Fisher forecasts for N_eff parameter.
 +
 +
DRAFT tool minimises variance in the CMB or Compton-y map by combining data from different frequency bands in an optimal way using internal linear combination. It also has the capacity to explicitly null foregrounds given a SED. More details about DRAFT can be found [https://docs.google.com/document/d/1Hy7QKOLmKBz3clOqA-P58weg7mUm1s89Ywuvlmgy9dU/edit here].
 +
 +
Link to DRAFT git repo: [https://github.com/sriniraghunathan/DRAFT DRAFT ILC tool].
 +
 +
Detailed report about N_eff Fisher forecasts using DRAFT can be found [https://docs.google.com/presentation/d/1_H93AMwg1uPjf7VcNFsffwhDo2XpNciXVV8NzWH9CnU/edit#slide=id.g865d806d9f_1_0 here].
 +
 +
==Inputs to DRAFT==
 +
*'''Noise power:''' 27, 39, 93, 145, 225 and 278 GHz noise curves for S4-wide. Specs obtained from this [https://cmb-s4.org/wiki/index.php/Expected_Survey_Performance_for_Science_Forecasting#Instrument_Definition link].
 +
 +
*'''Extragalactic foregrounds:''' Radio, CIB, tSZ and  kSZ power spectra from SPT measurements (George et al. 2015; arXiv: 1408.3161).
 +
 +
*'''Galactic foregrounds:''' Dust and Synchrotron power spectra obtained from pySM3 simulations.
 +
 +
*'''S4-LAT (default) footprint:''' fsky = 0.68 (0.77) using el=40 (30) scan strategy. We use el=40 scan strategy for all the forecasts here.
 +
 +
==DRAFT outputs==
 +
 +
* ILC noise curves.
 +
* Cosmological parameter constraints using Fisher forecasts: Specifically 𝞼(Neff) for different levels of galaxy masking (fsky) and for different S4-LAT configurations.
 +
 +
==Handling galactic emission==
 +
Since S4-LAT footprint is large (fsky = 0.68), the galactic emission must also be taken into account along with extragalactic foregrounds. However, we do not completely mask the regions by galactic emission. If we do, then we will miss the targeted sensitivity on the Neff parameter of 𝞼(Neff) = 0.03. So, we check the effect of galactic emission on 𝞼(Neff) by splitting the S4-LAT footprint into '''Clean''' and '''Dirty''' regions. The below figure illustrates three different '''Clean/Dirty''' masks and we use the rightmost as our fiducial mask.
 +
 +
[[File:S4_LAT_Neff_clean_dirt_masks.png | 600px]]
 +
 +
pySM3 simulations ([https://github.com/CMB-S4/s4mapbasedsims/tree/master/202002_foregrounds_extragalactic_cmb_tophat link]) were used to model the galactic foregrounds in the S4 patch. We compute the power spectra of galactic dust/synchrotron within each of the masks shown in the figure above and input that to DRAFT tool. As can be seen in this figure below, there is a range of \ells where the CMB E-mode power spectrum can be measured with SNR>1 even when we are directly looking through the plane of our galaxy (masks 3/4/5 in the figure) and this can help in reducing the error on Neff parameter. However, we have not yet quantified the bias in Neff due to galactic foregrounds and this is work in progress.
 +
 +
[[File:galactic_foreground_power_spectra.png | 600px]]
 +
 +
 +
==Fisher forecasting==
 +
* We use S4-like delensed CMB TT/EE/TE power spectra in the multiple range 30 <= \ell <= 5000 for Fisher forecasts. Check this [https://arxiv.org/abs/1609.08143 paper] for more information about delensing.
 +
* Lensing information is also added using lensing power spectrum calculated using both S4-like temperature and polarisation data. Since foregrounds can contaminate lensing, we restrict CMB multipole (small) \ell_max = 3000 for temperature based lensing reconstruction. For polarisation-based lensing estimators we set \ell_max = 5000.
 +
* Planck data is included in two different ways:
 +
** We use a Gaussian prior with width = 0.007 on tau parameter.
 +
** We include Planck CMB and lensing power spectra (2 <= \ell <= 2500) calculated in regions (fsky = 0.18) not covered by S4 footprint and not masked by Planck. Slide 4 in this [https://docs.google.com/presentation/d/1_H93AMwg1uPjf7VcNFsffwhDo2XpNciXVV8NzWH9CnU/edit#slide=id.g8b801fc060_5_0 report] give more details about this.
 +
* Fisher results are shown in the figure below.
 +
 +
[[File:DRAFT_results.png| 800px]]
 +
 +
==Pending items==
 +
* Quantify foreground bias on Neff parameter.
 +
* The masks shown above were not computed based on the actual level of galactic emission. So they must be replaced using Planck 857 GHz data.
 +
 +
 +
-------------------------------------------------------
 +
 +
==Some preliminary results/Postings prior to April 2020==
  
 
This posting compares Fisher forecasts for N_eff using the [https://github.com/sriniraghunathan/DRAFT DRAFT ILC tool].
 
This posting compares Fisher forecasts for N_eff using the [https://github.com/sriniraghunathan/DRAFT DRAFT ILC tool].
Line 259: Line 309:
 
=== Comparison of sigma(Neff) for S4-wide baseline configuration===
 
=== Comparison of sigma(Neff) for S4-wide baseline configuration===
  
The baseline configuration contains 2 LF (27/39 GHz) tubes, 12 MF (93/145 GHz) tubes, and 5 HF (225/278 GHz) tubes.
+
The baseline configuration contains 2 LF (27/39 GHz) tubes, 12 MF (93/145 GHz) tubes, and 5 HF (225/278 GHz) tubes. In the table below, S4 corresponds to S4-only CMB TTEETE; +Planck includes Planck data + tau prior along with the first column; + lensing includes S4 lensing power spectrum a+ BAO includes DESI BAO data along with other columns.
 +
 
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
Line 269: Line 320:
 
| 0.0456 (0.0456) || 0.0444 (0.0444) || 0.0368 (0.0375) || 0.0346 (0.0355) || 0.0387 (0.0387) || 0.0376 (0.0376) || 0.0361 (0.0364) || 0.0345 (0.0349) || 0.0368 (0.0368) || 0.0350 (0.0350) || 0.0331 (0.0334) || 0.0311 (0.0314) || style="text-align: center;" | BW  
 
| 0.0456 (0.0456) || 0.0444 (0.0444) || 0.0368 (0.0375) || 0.0346 (0.0355) || 0.0387 (0.0387) || 0.0376 (0.0376) || 0.0361 (0.0364) || 0.0345 (0.0349) || 0.0368 (0.0368) || 0.0350 (0.0350) || 0.0331 (0.0334) || 0.0311 (0.0314) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_L0 (withG) || JM_L1 (withG) || JM_L2 (withG) || JM_DL3 (withG) || JM_DL0 (withG) || JM_DL1 (withG) || JM_DL2 (withG) || JM_UL3 (withG) || JM_UL0 (withG) || JM_UL1 (withG) || JM_UL2 (withG) || JM_UL3 (withG) || style="text-align: center;" | JM
+
| style="color: green" | 0.0401 (0.0429) || style="color: green" | 0.0394 (0.0420) || style="color: green" | 0.0326 (0.0346) || style="color: green" | 0.0326 (0.0346) || style="color: green" | 0.0368 (0.0393) || style="color: green" | 0.0363 (0.0385) || style="color: green" | 0.0315 (0.0335) || style="color: green" | 0.0315 (0.0335) || style="color: green" | 0.0340 (0.0359) || style="color: green" | 0.0327 (0.0347) || style="color: green" | 0.0295 (0.0314) || style="color: green" | 0.0295 (0.0314) || style="text-align: center;" | JM
|-
 
| SR_L0 (withG) || SR_L1 (withG) || SR_L2 (withG) || SR_DL3 (withG) || SR_DL0 (withG) || SR_DL1 (withG) || SR_DL2 (withG) || SR_UL3 (withG) || SR_UL0 (withG) || SR_UL1 (withG) || SR_UL2 (withG) || SR_UL3 (withG) || style="text-align: center;" | SR
 
 
|-
 
|-
 +
| style="color: red" |  0.0417 (withG) || style="color: red" | 0.0413 (withG) || style="color: red" | 0.0345 (withG) || style="color: red" | SR_L3 (withG) || style="color: red" |  0.0375 (withG) || style="color: red" | 0.0370 (withG) || style="color: red" | 0.0346 (withG) || style="color: red" | SR_DL3 (withG) || style="color: red" |  0.0341 (withG) || style="color: red" | 0.0386 (withG) || style="color: red" | 0.0306 (withG) || style="color: red" | SR_UL3 (withG) || style="text-align: center;" | SR
 +
|----
 
| rowspan="3"| 0.59 (Mask 1)
 
| rowspan="3"| 0.59 (Mask 1)
 
| 0.0432 (0.0432) || 0.0427 (0.0427) || 0.0351 (0.0359) || 0.0332 (0.0341) || 0.0367 (0.0367) || 0.0360 (0.0360) || 0.0346 (0.0349) || 0.0331 (0.0335) || 0.0349 (0.0349) || 0.0335 (0.0335) || 0.0317 (0.0320) || 0.0298 (0.0302) || style="text-align: center;" | BW  
 
| 0.0432 (0.0432) || 0.0427 (0.0427) || 0.0351 (0.0359) || 0.0332 (0.0341) || 0.0367 (0.0367) || 0.0360 (0.0360) || 0.0346 (0.0349) || 0.0331 (0.0335) || 0.0349 (0.0349) || 0.0335 (0.0335) || 0.0317 (0.0320) || 0.0298 (0.0302) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_L0 (withG) || JM_L1 (withG) || JM_L2 (withG) || JM_DL3 (withG) || JM_DL0 (withG) || JM_DL1 (withG) || JM_DL2 (withG) || JM_UL3 (withG) || JM_UL0 (withG) || JM_UL1 (withG) || JM_UL2 (withG) || JM_UL3 (withG) || style="text-align: center;" | JM  
+
| style="color: green" | 0.0380 (0.0404) || style="color: green" | 0.0377 (0.0400) || style="color: green" | 0.0311 (0.0329) || style="color: green" | 0.0311 (0.0329) || style="color: green" | 0.0349 (0.0371) || style="color: green" | 0.0347 (0.0367) || style="color: green" | 0.0301 (0.0319) || style="color: green" | 0.0301 (0.0319) || style="color: green" | 0.0323 (0.0339) || style="color: green" | 0.0315 (0.0332) || style="color: green" | 0.0281 (0.0299) || style="color: green" | 0.0281 (0.0299) || style="text-align: center;" | JM  
 
|-
 
|-
| SR_L0 (withG) || SR_L1 (withG) || SR_L2 (withG) || SR_DL3 (withG) || SR_DL0 (withG) || SR_DL1 (withG) || SR_DL2 (withG) || SR_UL3 (withG) || SR_UL0 (withG) || SR_UL1 (withG) || SR_UL2 (withG) || SR_UL3 (withG) || style="text-align: center;" | SR  
+
| style="color: red" |  0.0396 (withG) || style="color: red" | 0.0392 (withG) || style="color: red" | 0.0327 (withG) || style="color: red" | SR_L3 (withG) || style="color: red" |  0.0356 (withG) || style="color: red" | 0.0351 (withG) || style="color: red" | 0.0328 (withG) || style="color: red" | SR_DL3 (withG) || style="color: red" |  0.0324 (withG) || style="color: red" | 0.0366 (withG) || style="color: red" | 0.0290 (withG) || style="color: red" | SR_UL3 (withG) || style="text-align: center;" | SR  
 
|-
 
|-
 
| rowspan="3"| 0.65 (Mask 2)
 
| rowspan="3"| 0.65 (Mask 2)
 
| 0.0411 (0.0411) || 0.0407 (0.0407) || 0.0335 (0.0344) || 0.0317 (0.0327) || 0.0350 (0.0350) || 0.0343 (0.0343) || 0.0330 (0.0333) || 0.0316 (0.0320) || 0.0332 (0.0332) || 0.0319 (0.0319) || 0.0302 (0.0305) || 0.0285 (0.0289) || style="text-align: center;" | BW  
 
| 0.0411 (0.0411) || 0.0407 (0.0407) || 0.0335 (0.0344) || 0.0317 (0.0327) || 0.0350 (0.0350) || 0.0343 (0.0343) || 0.0330 (0.0333) || 0.0316 (0.0320) || 0.0332 (0.0332) || 0.0319 (0.0319) || 0.0302 (0.0305) || 0.0285 (0.0289) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_L0 (withG) || JM_L1 (withG) || JM_L2 (withG) || JM_DL3 (withG) || JM_DL0 (withG) || JM_DL1 (withG) || JM_DL2 (withG) || JM_UL3 (withG) || JM_UL0 (withG) || JM_UL1 (withG) || JM_UL2 (withG) || JM_UL3 (withG) || style="text-align: center;" | JM  
+
| style="color: green" | 0.0362 (0.0388) || style="color: green" | 0.0360 (0.0384) || style="color: green" | 0.0297 (0.0317) || style="color: green" | 0.0297 (0.0317) || style="color: green" | 0.0332 (0.0371) || style="color: green" | 0.0331 (0.0353) || style="color: green" | 0.0287 (0.0307) || style="color: green" | 0.0287 (0.0307) || style="color: green" | 0.0307 (0.0325) || style="color: green" | 0.0300 (0.0318) || style="color: green" | 0.0268 (0.0287) || style="color: green" | 0.0268 (0.0287) || style="text-align: center;" | JM  
 
|-
 
|-
| SR_L0 (withG) || SR_L1 (withG) || SR_L2 (withG) || SR_DL3 (withG) || SR_DL0 (withG) || SR_DL1 (withG) || SR_DL2 (withG) || SR_UL3 (withG) || SR_UL0 (withG) || SR_UL1 (withG) || SR_UL2 (withG) || SR_UL3 (withG) || style="text-align: center;" | SR  
+
| style="color: red" |  0.0377 (withG) || style="color: red" | 0.0373 (withG) || style="color: red" | 0.0311 (withG) || style="color: red" | SR_L3 (withG) || style="color: red" |  0.0339 (withG) || style="color: red" | 0.0334 (withG) || style="color: red" | 0.0312 (withG) || style="color: red" | SR_DL3 (withG) || style="color: red" |  0.0308 (withG) || style="color: red" | 0.0348 (withG) || style="color: red" | 0.0276 (withG) || style="color: red" | SR_UL3 (withG) || style="text-align: center;" | SR  
 
|-
 
|-
 
| rowspan="3"| 0.77 (Mask 3)
 
| rowspan="3"| 0.77 (Mask 3)
 
| 0.0377 (0.0377) || 0.0374 (0.0374) || 0.0307 (0.0319) || 0.0292 (0.0306) || 0.0321 (0.0321) || 0.0316 (0.0316) || 0.0303 (0.0307) || 0.0291 (0.0296) || 0.0305 (0.0305) || 0.0293 (0.0293) || 0.0277 (0.0281) || 0.0263 (0.0268) || style="text-align: center;" | BW  
 
| 0.0377 (0.0377) || 0.0374 (0.0374) || 0.0307 (0.0319) || 0.0292 (0.0306) || 0.0321 (0.0321) || 0.0316 (0.0316) || 0.0303 (0.0307) || 0.0291 (0.0296) || 0.0305 (0.0305) || 0.0293 (0.0293) || 0.0277 (0.0281) || 0.0263 (0.0268) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_L0 (withG) || JM_L1 (withG) || JM_L2 (withG) || JM_DL3 (withG) || JM_DL0 (withG) || JM_DL1 (withG) || JM_DL2 (withG) || JM_UL3 (withG) || JM_UL0 (withG) || JM_UL1 (withG) || JM_UL2 (withG) || JM_UL3 (withG) || style="text-align: center;" | JM  
+
| style="color: green" | 0.0332 (0.0369) || style="color: green" | 0.0331 (0.0365) || style="color: green" | 0.0273 (0.0301) || style="color: green" | 0.0273 (0.0301) || style="color: green" | 0.0305 (0.0339) || style="color: green" | 0.0305 (0.0335) || style="color: green" | 0.0264 (0.0292) || style="color: green" | 0.0264 (0.0292) || style="color: green" | 0.0282 (0.0307) || style="color: green" | 0.0276 (0.0300) || style="color: green" | 0.0247 (0.0272) || style="color: green" | 0.0247 (0.0272) || style="text-align: center;" | JM  
 
|-
 
|-
| SR_L0 (withG) || SR_L1 (withG) || SR_L2 (withG) || SR_DL3 (withG) || SR_DL0 (withG) || SR_DL1 (withG) || SR_DL2 (withG) || SR_UL3 (withG) || SR_UL0 (withG) || SR_UL1 (withG) || SR_UL2 (withG) || SR_UL3 (withG) || style="text-align: center;" | SR  
+
| style="color: red" |  0.0345 (withG) || style="color: red" | 0.0342 (withG) || style="color: red" | 0.0285 (withG) || style="color: red" | SR_L3 (withG) || style="color: red" |  0.0311 (withG) || style="color: red" | 0.0306 (withG) || style="color: red" | 0.0286 (withG) || style="color: red" | SR_DL3 (withG) || style="color: red" |  0.0283 (withG) || style="color: red" | 0.0319 (withG) || style="color: red" | 0.0253 (withG) || style="color: red" | SR_UL3 (withG) || style="text-align: center;" | SR  
 
|-
 
|-
 
|}
 
|}
  
 
BW preliminary so far.
 
BW preliminary so far.
 +
 +
JM rows: S4-Only numbers have m_nu fixed, while others marginalize over m_nu.  l_max^TT = 5000 for all results, which seems to account for discrepancy with BW and SR.
 +
 +
[[File:JM_BW_gal0.png | 400px]]
 +
[[File:JM_SR_gal0.png | 400px]]
 +
[[File:BW_SR_gal0.png | 400px]]
  
 
=== Comparison of sigma(Neff) between S4-wide baseline and other configurations ===
 
=== Comparison of sigma(Neff) between S4-wide baseline and other configurations ===
Line 309: Line 366:
 
| 0.0345 (0.0349) || 0.0345 (withG) || 0.0344 (withG) || 0.0343 (withG) || 0.0344 (withG) || 0.0345 (withG) || style="text-align: center;" | BW  
 
| 0.0345 (0.0349) || 0.0345 (withG) || 0.0344 (withG) || 0.0343 (withG) || 0.0344 (withG) || 0.0345 (withG) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_DL_b (withG) || JM_DL_c1 (withG) || JM_DL_c2 (withG) || JM_DL_c3 (withG) || JM_DL_c4 (withG) || JM_DL_c5 (withG)|| style="text-align: center;" | JM  
+
| 0.0315 (0.0335) || 0.0315 (withG) || 0.0314 (withG) || 0.0313 (withG) || 0.0315 (withG) || 0.0317 (withG)|| style="text-align: center;" | JM  
 
|-
 
|-
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
Line 316: Line 373:
 
| 0.0331 (0.0335) || 0.0330 (withG) || 0.0330 (withG) || 0.0329 (withG) || 0.033 (withG) || 0.0331 (withG) || style="text-align: center;" | BW  
 
| 0.0331 (0.0335) || 0.0330 (withG) || 0.0330 (withG) || 0.0329 (withG) || 0.033 (withG) || 0.0331 (withG) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_DL_b (withG) || JM_DL_c1 (withG) || JM_DL_c2 (withG) || JM_DL_c3 (withG) || JM_DL_c4 (withG) || JM_DL_c5 (withG)|| style="text-align: center;" | JM  
+
| 0.0301 (0.0319) || 0.0300 (withG) || 0.0300 (withG) || 0.0299 (withG) || 0.0301 (withG) || 0.0302 (withG)|| style="text-align: center;" | JM  
 
|-
 
|-
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
Line 323: Line 380:
 
| 0.0316 (0.0320) || 0.0316 (withG) || 0.0315 (withG) || 0.0315 (withG) || 0.0315 (withG) || 0.0316 (withG) || style="text-align: center;" | BW  
 
| 0.0316 (0.0320) || 0.0316 (withG) || 0.0315 (withG) || 0.0315 (withG) || 0.0315 (withG) || 0.0316 (withG) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_DL_b (withG) || JM_DL_c1 (withG) || JM_DL_c2 (withG) || JM_DL_c3 (withG) || JM_DL_c4 (withG) || JM_DL_c5 (withG)|| style="text-align: center;" | JM  
+
| 0.0287 (0.0307) || 0.0287 (withG) || 0.0286 (withG) || 0.0286 (withG) || 0.0287 (withG) || 0.0289 (withG)|| style="text-align: center;" | JM  
 
|-
 
|-
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
Line 330: Line 387:
 
| 0.0291 (0.0296) || 0.0291 (withG) || 0.0290 (withG) || 0.0290 (withG) || 0.0291 (withG) || 0.0291 (withG) || style="text-align: center;" | BW  
 
| 0.0291 (0.0296) || 0.0291 (withG) || 0.0290 (withG) || 0.0290 (withG) || 0.0291 (withG) || 0.0291 (withG) || style="text-align: center;" | BW  
 
|-
 
|-
| JM_DL_b (withG) || JM_DL_c1 (withG) || JM_DL_c2 (withG) || JM_DL_c3 (withG) || JM_DL_c4 (withG) || JM_DL_c5 (withG)|| style="text-align: center;" | JM  
+
| 0.0264 (0.0292) || 0.0264 (withG) || 0.0264 (withG) || 0.0263 (withG) || 0.0264 (withG) || 0.0266 (withG)|| style="text-align: center;" | JM  
 
|-
 
|-
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
 
| SR_DL_b (withG) || SR_DL_c1 (withG) || SR_DL_c2 (withG) || SR_DL_c3 (withG) || SR_DL_c4 (withG) || SR_DL_c5 (withG)|| style="text-align: center;" | SR  
 
|-
 
|-
 
|}
 
|}

Latest revision as of 12:58, 12 October 2020

Overview

This posting summarises the DRAFT (Dark Radiation Anisotropy Flowdown Team) tool and the Fisher forecasts for N_eff parameter.

DRAFT tool minimises variance in the CMB or Compton-y map by combining data from different frequency bands in an optimal way using internal linear combination. It also has the capacity to explicitly null foregrounds given a SED. More details about DRAFT can be found here.

Link to DRAFT git repo: DRAFT ILC tool.

Detailed report about N_eff Fisher forecasts using DRAFT can be found here.

Inputs to DRAFT

  • Noise power: 27, 39, 93, 145, 225 and 278 GHz noise curves for S4-wide. Specs obtained from this link.
  • Extragalactic foregrounds: Radio, CIB, tSZ and kSZ power spectra from SPT measurements (George et al. 2015; arXiv: 1408.3161).
  • Galactic foregrounds: Dust and Synchrotron power spectra obtained from pySM3 simulations.
  • S4-LAT (default) footprint: fsky = 0.68 (0.77) using el=40 (30) scan strategy. We use el=40 scan strategy for all the forecasts here.

DRAFT outputs

  • ILC noise curves.
  • Cosmological parameter constraints using Fisher forecasts: Specifically 𝞼(Neff) for different levels of galaxy masking (fsky) and for different S4-LAT configurations.

Handling galactic emission

Since S4-LAT footprint is large (fsky = 0.68), the galactic emission must also be taken into account along with extragalactic foregrounds. However, we do not completely mask the regions by galactic emission. If we do, then we will miss the targeted sensitivity on the Neff parameter of 𝞼(Neff) = 0.03. So, we check the effect of galactic emission on 𝞼(Neff) by splitting the S4-LAT footprint into Clean and Dirty regions. The below figure illustrates three different Clean/Dirty masks and we use the rightmost as our fiducial mask.

S4 LAT Neff clean dirt masks.png

pySM3 simulations (link) were used to model the galactic foregrounds in the S4 patch. We compute the power spectra of galactic dust/synchrotron within each of the masks shown in the figure above and input that to DRAFT tool. As can be seen in this figure below, there is a range of \ells where the CMB E-mode power spectrum can be measured with SNR>1 even when we are directly looking through the plane of our galaxy (masks 3/4/5 in the figure) and this can help in reducing the error on Neff parameter. However, we have not yet quantified the bias in Neff due to galactic foregrounds and this is work in progress.

Galactic foreground power spectra.png


Fisher forecasting

  • We use S4-like delensed CMB TT/EE/TE power spectra in the multiple range 30 <= \ell <= 5000 for Fisher forecasts. Check this paper for more information about delensing.
  • Lensing information is also added using lensing power spectrum calculated using both S4-like temperature and polarisation data. Since foregrounds can contaminate lensing, we restrict CMB multipole (small) \ell_max = 3000 for temperature based lensing reconstruction. For polarisation-based lensing estimators we set \ell_max = 5000.
  • Planck data is included in two different ways:
    • We use a Gaussian prior with width = 0.007 on tau parameter.
    • We include Planck CMB and lensing power spectra (2 <= \ell <= 2500) calculated in regions (fsky = 0.18) not covered by S4 footprint and not masked by Planck. Slide 4 in this report give more details about this.
  • Fisher results are shown in the figure below.

DRAFT results.png

Pending items

  • Quantify foreground bias on Neff parameter.
  • The masks shown above were not computed based on the actual level of galactic emission. So they must be replaced using Planck 857 GHz data.



Some preliminary results/Postings prior to April 2020

This posting compares Fisher forecasts for N_eff using the DRAFT ILC tool.


Results from Joel

Joel's fisher forecasting code was used in the WAFTT study (Noise models and sky fractions for WAFTT, WAFTT results part 2, WAFTT results part 3).

We assume l_min = 30 for TT, TE, and EE spectra, and l_max = 5000 for all CMB-S4 spectra. Planck is included in addition to CMB-S4 in a few non-trivial ways. First, Planck is added with inverse variance weighting to the the CMB-S4 ILC, which provides a significant reduction in noise on large angular scales (due to atmospheric contamination of CMB-S4 observations). Second, a prior of \sigma(\tau) = 0.007, derived from Planck EE measurements is included. Third, Planck is assumed to cover an additional patch of sky not observed by CMB-S4, up to a maximum of 60 percent of the sky (this is a slightly conservative choice, and for the scan strategies considered here typically means no additional sky coverage from Planck). Last, Planck measurements of large angular scale temperature is included. Specifically, CMB-S4 is assumed to have l_min = 30 for T and E, but Planck provides measurements of TT between l=2 and l=30 across 80 percent of the sky.

The code computes CMB spectra using CAMB. Cosmological parameters, fiducial values, and step sizes are shown in the table below.

Parameter Fiducial Value Step Size
omega_c_h2 0.1197 0.0030
omega_b_h2 0.0222 0.0008
N_eff 3.046 0.080
A_s 2.196e-9 0.1e-9
n_s 0.9655 0.010
tau 0.060 0.020
theta_s 0.010409 0.000050
mnu 0.06 0.02

The primordial helium abundance Y_p is set to be consistent with the predictions of standard Big Bang Nucleosynthesis (such that Y_p is a function of N_eff and \omega_b h^2). As will be shown below fixing Y_p to a constant or marginalizing over a freely varying Y_p increases errors on N_eff.

Lensing reconstruction noise is estimated with the minimum variance combination of the quadratic estimators, including the improvement which comes from iterative EB reconstruction. Delensing is performed using the prescription of CMB Delensing Beyond the B Modes.

Information from DESI BAO is included, though it has little impact on N_eff errors. BAO plays a significant role in the errors for neutrino mass, however.

Mask Lensed Delensed Unlensed
galmask0 0.0346 0.0335 0.0314
galmask1 0.0329 0.0319 0.0299
galmask2 0.0317 0.0307 0.0287
galmask3 0.0301 0.0292 0.0272
nogalmask0 0.0326 0.0315 0.0295
nogalmask1 0.0311 0.0301 0.0281
nogalmask2 0.0297 0.0287 0.0268
nogalmask3 0.0273 0.0264 0.0247


We will now give results for a few different forecasting choices for the sake of comparisons with other codes.

No lensing reconstruction:

Mask Lensed Delensed Unlensed
galmask0 0.0420 0.0385 0.0347
galmask1 0.0400 0.0367 0.0332
galmask2 0.0384 0.0353 0.0318
galmask3 0.0365 0.0335 0.0300
nogalmask3 0.0330 0.0304 0.0275


With l_max^TT = 3000:

Mask Lensed Delensed Unlensed
galmask0 0.0357 0.0347 0.0333
galmask1 0.0340 0.0331 0.0318
galmask2 0.0327 0.0318 0.0305
galmask3 0.0311 0.0302 0.0289
nogalmask3 0.0284 0.0276 0.0266

Excluding DESI BAO changes N_eff errors only in the last decimal place (at the level of 0.0001).

Fixing Y_p to a constant (rather than setting by BBN consistency):

Mask Lensed Delensed Unlensed
galmask0 0.0425 0.0409 0.0383
galmask1 0.0405 0.0390 0.0365
galmask2 0.0389 0.0375 0.0350
galmask3 0.0370 0.0356 0.0332
nogalmask3 0.0336 0.0323 0.0302

Marginalizing over Y_p:

Mask Lensed Delensed Unlensed
galmask0 0.0967 0.0866 0.0804
galmask1 0.0929 0.0832 0.0772
galmask2 0.0890 0.0797 0.0739
galmask3 0.0831 0.0747 0.0692
nogalmask3 0.0792 0.0704 0.0656




Comparison of sigma(Neff) for S4-wide baseline configuration

The baseline configuration contains 2 LF (27/39 GHz) tubes, 12 MF (93/145 GHz) tubes, and 5 HF (225/278 GHz) tubes. In the table below, S4 corresponds to S4-only CMB TTEETE; +Planck includes Planck data + tau prior along with the first column; + lensing includes S4 lensing power spectrum a+ BAO includes DESI BAO data along with other columns.

Fsky Lensed: No galaxy (with galaxy) Delensed (Alens = 0.3): No galaxy (with galaxy) Unlensed: No galaxy (with galaxy) Author
S4 (TTTEEE) +Planck + Lensing + BAO S4 (TTTEEE) +Planck + Lensing + BAO S4 (TTTEEE) +Planck + Lensing + BAO
0.53 (Mask 0) 0.0456 (0.0456) 0.0444 (0.0444) 0.0368 (0.0375) 0.0346 (0.0355) 0.0387 (0.0387) 0.0376 (0.0376) 0.0361 (0.0364) 0.0345 (0.0349) 0.0368 (0.0368) 0.0350 (0.0350) 0.0331 (0.0334) 0.0311 (0.0314) BW
0.0401 (0.0429) 0.0394 (0.0420) 0.0326 (0.0346) 0.0326 (0.0346) 0.0368 (0.0393) 0.0363 (0.0385) 0.0315 (0.0335) 0.0315 (0.0335) 0.0340 (0.0359) 0.0327 (0.0347) 0.0295 (0.0314) 0.0295 (0.0314) JM
0.0417 (withG) 0.0413 (withG) 0.0345 (withG) SR_L3 (withG) 0.0375 (withG) 0.0370 (withG) 0.0346 (withG) SR_DL3 (withG) 0.0341 (withG) 0.0386 (withG) 0.0306 (withG) SR_UL3 (withG) SR
0.59 (Mask 1) 0.0432 (0.0432) 0.0427 (0.0427) 0.0351 (0.0359) 0.0332 (0.0341) 0.0367 (0.0367) 0.0360 (0.0360) 0.0346 (0.0349) 0.0331 (0.0335) 0.0349 (0.0349) 0.0335 (0.0335) 0.0317 (0.0320) 0.0298 (0.0302) BW
0.0380 (0.0404) 0.0377 (0.0400) 0.0311 (0.0329) 0.0311 (0.0329) 0.0349 (0.0371) 0.0347 (0.0367) 0.0301 (0.0319) 0.0301 (0.0319) 0.0323 (0.0339) 0.0315 (0.0332) 0.0281 (0.0299) 0.0281 (0.0299) JM
0.0396 (withG) 0.0392 (withG) 0.0327 (withG) SR_L3 (withG) 0.0356 (withG) 0.0351 (withG) 0.0328 (withG) SR_DL3 (withG) 0.0324 (withG) 0.0366 (withG) 0.0290 (withG) SR_UL3 (withG) SR
0.65 (Mask 2) 0.0411 (0.0411) 0.0407 (0.0407) 0.0335 (0.0344) 0.0317 (0.0327) 0.0350 (0.0350) 0.0343 (0.0343) 0.0330 (0.0333) 0.0316 (0.0320) 0.0332 (0.0332) 0.0319 (0.0319) 0.0302 (0.0305) 0.0285 (0.0289) BW
0.0362 (0.0388) 0.0360 (0.0384) 0.0297 (0.0317) 0.0297 (0.0317) 0.0332 (0.0371) 0.0331 (0.0353) 0.0287 (0.0307) 0.0287 (0.0307) 0.0307 (0.0325) 0.0300 (0.0318) 0.0268 (0.0287) 0.0268 (0.0287) JM
0.0377 (withG) 0.0373 (withG) 0.0311 (withG) SR_L3 (withG) 0.0339 (withG) 0.0334 (withG) 0.0312 (withG) SR_DL3 (withG) 0.0308 (withG) 0.0348 (withG) 0.0276 (withG) SR_UL3 (withG) SR
0.77 (Mask 3) 0.0377 (0.0377) 0.0374 (0.0374) 0.0307 (0.0319) 0.0292 (0.0306) 0.0321 (0.0321) 0.0316 (0.0316) 0.0303 (0.0307) 0.0291 (0.0296) 0.0305 (0.0305) 0.0293 (0.0293) 0.0277 (0.0281) 0.0263 (0.0268) BW
0.0332 (0.0369) 0.0331 (0.0365) 0.0273 (0.0301) 0.0273 (0.0301) 0.0305 (0.0339) 0.0305 (0.0335) 0.0264 (0.0292) 0.0264 (0.0292) 0.0282 (0.0307) 0.0276 (0.0300) 0.0247 (0.0272) 0.0247 (0.0272) JM
0.0345 (withG) 0.0342 (withG) 0.0285 (withG) SR_L3 (withG) 0.0311 (withG) 0.0306 (withG) 0.0286 (withG) SR_DL3 (withG) 0.0283 (withG) 0.0319 (withG) 0.0253 (withG) SR_UL3 (withG) SR

BW preliminary so far.

JM rows: S4-Only numbers have m_nu fixed, while others marginalize over m_nu. l_max^TT = 5000 for all results, which seems to account for discrepancy with BW and SR.

JM BW gal0.png JM SR gal0.png BW SR gal0.png

Comparison of sigma(Neff) between S4-wide baseline and other configurations

These are computed for Delensed case (Alens = 0.3) and contains information from S4 + Planck + lensing + BAO

Fsky Baseline 2 LF/12 MF/5 HF: No (with) galaxy 1 LF/12 MF/6 HF: No (with) galaxy 1 LF/13 MF/5 HF: No (with) galaxy 1 LF/14 MF/4 HF: No (with) galaxy 2 LF/13 MF/4 HF: No (with) galaxy 3 LF/12 MF/4 HF: No (with) galaxy Authour
0.53 (Mask 0) 0.0345 (0.0349) 0.0345 (withG) 0.0344 (withG) 0.0343 (withG) 0.0344 (withG) 0.0345 (withG) BW
0.0315 (0.0335) 0.0315 (withG) 0.0314 (withG) 0.0313 (withG) 0.0315 (withG) 0.0317 (withG) JM
SR_DL_b (withG) SR_DL_c1 (withG) SR_DL_c2 (withG) SR_DL_c3 (withG) SR_DL_c4 (withG) SR_DL_c5 (withG) SR
0.59 (Mask 1) 0.0331 (0.0335) 0.0330 (withG) 0.0330 (withG) 0.0329 (withG) 0.033 (withG) 0.0331 (withG) BW
0.0301 (0.0319) 0.0300 (withG) 0.0300 (withG) 0.0299 (withG) 0.0301 (withG) 0.0302 (withG) JM
SR_DL_b (withG) SR_DL_c1 (withG) SR_DL_c2 (withG) SR_DL_c3 (withG) SR_DL_c4 (withG) SR_DL_c5 (withG) SR
0.65 (Mask 2) 0.0316 (0.0320) 0.0316 (withG) 0.0315 (withG) 0.0315 (withG) 0.0315 (withG) 0.0316 (withG) BW
0.0287 (0.0307) 0.0287 (withG) 0.0286 (withG) 0.0286 (withG) 0.0287 (withG) 0.0289 (withG) JM
SR_DL_b (withG) SR_DL_c1 (withG) SR_DL_c2 (withG) SR_DL_c3 (withG) SR_DL_c4 (withG) SR_DL_c5 (withG) SR
0.77 (Mask 3) 0.0291 (0.0296) 0.0291 (withG) 0.0290 (withG) 0.0290 (withG) 0.0291 (withG) 0.0291 (withG) BW
0.0264 (0.0292) 0.0264 (withG) 0.0264 (withG) 0.0263 (withG) 0.0264 (withG) 0.0266 (withG) JM
SR_DL_b (withG) SR_DL_c1 (withG) SR_DL_c2 (withG) SR_DL_c3 (withG) SR_DL_c4 (withG) SR_DL_c5 (withG) SR