Difference between revisions of "Harvard-2017:S2"

From CMB-S4 wiki
Jump to: navigation, search
(Notes from session)
(Notes from session)
 
Line 60: Line 60:
  
 
1) tau from B and phi.
 
1) tau from B and phi.
we want to measure tau because neutrino mass will be tau-limited.
+
We want to measure tau because neutrino mass will be tau-limited.
cv: sigma(tau)=0.002
+
CV: sigma(tau)=0.002
planck:sigma(tau)=0.006
+
 
 +
Planck:sigma(tau)=0.006
  
 
Reconstruct E-modes with B-modes and phi.
 
Reconstruct E-modes with B-modes and phi.
 +
 
S4 will have great maps of phi and of B.
 
S4 will have great maps of phi and of B.
We can tink about reconstructing low-ell E-modes.
+
We can think about reconstructing low-ell E-modes.
You need a very low noise CMB survey. You can cross-correlate this with E-mode measurements
+
You need a very low noise CMB survey. You can cross-correlate this with E-mode measurements.
To be competitive with Planck, we need Delta_T=a few times 10^{-2} muK-arcmin.
+
To be competitive with Planck, we need Delta_T=a few times 10^{-2} muK-arcmi (1.5 orders of magnitude lower noise than CMB-S4 noise).
 
Information coming from l_Bmodes~1000.
 
Information coming from l_Bmodes~1000.
Tnformation coming from l_phi~1000.
+
Information coming from l_phi~1000.
Information coming from squeezed triangles.
+
Information coming from squeezed triangles -> low-ell E-modes.
  
 
2) tau from polarized SZ effect:
 
2) tau from polarized SZ effect:
Line 79: Line 81:
 
The CV comes from the fluctuations of the quadrupoles that the electrons see.
 
The CV comes from the fluctuations of the quadrupoles that the electrons see.
  
If we get maps of E and maps of the quadrupole field, we can reconstruct the visibility function and therefore get tau without CV.
+
If we get maps of E-modes and maps of the quadrupole field, we can reconstruct the visibility function and therefore get tau without CV.
 
The quadrupole field is coherent.
 
The quadrupole field is coherent.
 
z=0 is about 50 percent correlated at redshift 1-2.
 
z=0 is about 50 percent correlated at redshift 1-2.
sigma(tau) at z=2 is 10 times below CV with CV limited measurement of tau and of E-modes.
+
sigma(tau) at z=2 is 10 times below CV with a CV limited measurement of E-modes and a future galaxy survey (DESI?).
  
 
-------------------------------------------------------------
 
-------------------------------------------------------------
Line 88: Line 90:
 
'''Georges Obied's talk:'''
 
'''Georges Obied's talk:'''
  
Transfer function: NO ISW effect for polarization vs. temperature (this is why the mappting between k and ell is much cleaner).
+
Transfer function: NO ISW effect for polarization vs. temperature (this is why the mapping between k and ell is much cleaner).
This cleans mapping means that the response of the power spectrum in ClEE is more dramatic than CLTT to Omegach2 (especially at ell~100-200).
+
This cleaner mapping means that the response of the power spectrum in ClEE is more dramatic than ClTT to Omegach2 (especially at ell~100-200).
 
Omegach2 is anti-correlated with H0.
 
Omegach2 is anti-correlated with H0.
Planck TE at ell>1000 has no constraining power. A better measurement in TE would help.
+
The intrinsic sensitivity of low-l (l<~200) E mode data to Omega_ch^2 (and therefore to H0) is higher in comparison to the temperature sensitivity.
 
+
Current polarization data also slightly enhance the significance of a sharp suppression of large-scale power but leave room for large improvements in the future with cosmic variance limited E-mode measurements. The suppression in power at low-ell is responsible for a higher value of H0 from the temperature ell<1000 in Planck.
Probes of:
+
Dipolar asymmetry.
+
  
 
== Action items/Next steps ==
 
== Action items/Next steps ==
  
 
Summarize action items here
 
Summarize action items here

Latest revision as of 09:51, 25 August 2017

Back to Harvard-2017 main page

Parallel Session S2: E-mode cosmology: open questions after Planck, H0 and other tensions, reionization (Chair: Cora Dvorkin) [Jefferson 256]

Post talks here.

Notes from session

Cora Dvorkin's talk:

Inflationary potential with E-modes


Usual procedure: Start with a particular potential, equations of motion, then compute curvature => get T and polarization power spectra.

Reverse engineering: polarization is a better probe of the features in the potential.

Transfer function of temperature and polarization: cleaner mapping of the modes from polarization. It makes it a better probe of features in the inflationary potential. This holds in principle at all scales.


With Planck, you can test features at ell 20-40 at the 2-3 sigma level. With cosmic variance-limited experiments, you should be able to test at the 5-8 sigma level. (Again because of differences in the transfer function.)


Polarization as a test of single-field inflation. Temperature provides 1 dof => prediction for E-mode power spectrum. E-modes provide a test for single-field inflation; could in principle rule such models out.


Q: What scales contribute the most to this test? ell < 10 seem to have a lot of scatter. Cora: all scales matter, it depends where you see the deviation. It’s a general tool to be used for anything outside of single-field inflation.

Q: Difficulty of following up deviations at low ell because of cosmic variance? Cora: at ell 20-40, with a measurement at better than 40% uncertainty you can already say something useful. Galaxy clustering is also a tracer of matter fluctuation.

Q: Has this technique been tested with Planck or Bicep data? Cora: Yes, the potential has been reconstructed with Planck data, not Bicep.

Q: Can you do this reconstruction with Bicep (concern of wider bands)? Features are narrow so need narrow bins.



Reionization from E-modes (Cora presenting on behalf of Vinicius Miranda, UPenn):


Other things you can do with E-modes: reconstructing tau.

Using latest Planck data, putting constraints on latest ionization history. No longer assume instantaneous 2-parameter ionization model (tau and width of history).

Let’s assume we have any function for the ionization history (no theoretical priors). Use a principal component method instead. Best-measured component is average optical depth.

Difference between 2 approaches with Planck 2015 data (HFI data will be interesting later): Optical depth at z = 10-15 is 2 sigma different.

They also obtain Tau at z of 15 = 0.033. Poor pinning down of tau is a major source of error in neutrino mass constraints. If this measurement persists in Planck-HFI data, then they won’t be able to get tau better than comic-variance limited from 21-cm surveys. This would be the consequence of such a large value of optical depth at high redshift. This would make tau probes from CMB really necessary.

Q: tau = 0.033 number is to interpreted as integrated from z = infinity to 15

Q: ground-based experiments to lower ell => see poster


Alex van Engelen's talk:

1) tau from B and phi. We want to measure tau because neutrino mass will be tau-limited. CV: sigma(tau)=0.002

Planck:sigma(tau)=0.006

Reconstruct E-modes with B-modes and phi.

S4 will have great maps of phi and of B. We can think about reconstructing low-ell E-modes. You need a very low noise CMB survey. You can cross-correlate this with E-mode measurements. To be competitive with Planck, we need Delta_T=a few times 10^{-2} muK-arcmi (1.5 orders of magnitude lower noise than CMB-S4 noise). Information coming from l_Bmodes~1000. Information coming from l_phi~1000. Information coming from squeezed triangles -> low-ell E-modes.

2) tau from polarized SZ effect: The polarized SZ effect comes from low redshift objects. You can see the CMB surface that electrons close to us see. Idea: to measure tau from reionization. E-modes that we measure come from a line of sight integral over the field of quadrupoles times the visibility function. The CV comes from the fluctuations of the quadrupoles that the electrons see.

If we get maps of E-modes and maps of the quadrupole field, we can reconstruct the visibility function and therefore get tau without CV. The quadrupole field is coherent. z=0 is about 50 percent correlated at redshift 1-2. sigma(tau) at z=2 is 10 times below CV with a CV limited measurement of E-modes and a future galaxy survey (DESI?).


Georges Obied's talk:

Transfer function: NO ISW effect for polarization vs. temperature (this is why the mapping between k and ell is much cleaner). This cleaner mapping means that the response of the power spectrum in ClEE is more dramatic than ClTT to Omegach2 (especially at ell~100-200). Omegach2 is anti-correlated with H0. The intrinsic sensitivity of low-l (l<~200) E mode data to Omega_ch^2 (and therefore to H0) is higher in comparison to the temperature sensitivity. Current polarization data also slightly enhance the significance of a sharp suppression of large-scale power but leave room for large improvements in the future with cosmic variance limited E-mode measurements. The suppression in power at low-ell is responsible for a higher value of H0 from the temperature ell<1000 in Planck.

Action items/Next steps

Summarize action items here