Probing Inflation and Reionization with Large-Scale CMB Polarization - - PowerPoint PPT Presentation

Vinícius Miranda

Department of Physics and Astronomy University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

Berkeley, December 4 2017 - Vinícius Miranda

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Part I - The Epoch Of Reionization First stars: source of ionizing radiation Difficult to model: radiative transfer + big volume One of the least understood aspects of cosmology

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

How Can We Probe The EOR? 21 cm Cosmic Microwave Background

Measure more than total optical depth Limitations: Cosmic Variance Better than Cosmic Variance Atmosphere limits redshift range

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Hypothesis: The Instantaneous Reionization Model Higher optical depth implies transition at higher redshift

Helium

τ(z1,z2) ∝ dz (1+ z)2 H(z) / H0 xe(z)

z1 z2

∫

The only free parameter given tau is the transition width.

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Hypothesis: The Instantaneous Reionization Model Low and high redshift behavior are linked together higher redshift => higher l

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Beyond IRM: Principal Components Analysis Eigenvectors covariance

−2 −1 1 2 3 4

Sa(z)

a = 1 a = 1 a = 2 a = 2 a = 3 a = 3 a = 4 a = 4 a = 5 a = 5 10 15 20 25 30

z

−0.1 0.0 0.1 0.2 0.3

τ(z, zmax)

xe(z) = xe

fid(z)+

ma

a

∑

Sa(z)

Lower PCs: better constrained a=1 ~ average optical depth a=2 ~ Difference high-low z

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

PCA: Completeness 5 PCs are complete! Complete in polarization error < cosmic variance

5 10 15 20 25 30 35 40 l 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 l(l + 1)CEE

l

/2π [µK2] tanh ML tanh ML PC PC ML

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Wayne Hu Chen He V . Miranda PCA Results On The EOR @ JPL @ Job market Has tenure

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

PCA Results On The EOR Is IRM favored by the data?

−0.1 0.0 0.1 0.2

m1

−0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

m2

Gaussian unphysical

xe(z) = xe

fid(z)+

ma

a

∑

Sa(z)

σ (m1) < σ (m2)< σ (m3) < ...

Instantaneous reionization

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Does the IRM spuriously disfavor high redshift sources?

5 10 15 20 25 30

z

−0.04 −0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14

τ(z, zmax)

tanh ML PC PC mean PC 68, 95% CL Planck 2015

τ PC(z = 15,zmax) = 0.033

PCA Results On The EOR

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

What the polarization spectrum look like?

5 10 15 20 25 30 35 40 l 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 l(l + 1)CEE

l

/2π [µK2] tanh ML tanh ML PC PC ML

Presence of high redshift sources does NOT imply unreasonable tau PCA Results On The EOR

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Forward Modeling Only Wayne Hu Chen He V . Miranda Adam Lidz Can models with metal-free stars be the source of the high redshift signal? PCA: Completeness

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

We solve the following ODE

Tanh Pop-II Pop-III Pop-III, self-regulated

d〈xi〉 dt = d dt ζ II fc,II +ζ III fc,III

( )− 〈xi〉

trec(z)

Baseline model (Pop-II stars) Efficiency

〈xT 〉 ≠ 〈xII 〉 + 〈xIII 〉

zend = 6

Source Of High Redshift Ionization: Pop-III?

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

PC Projection

Tanh P15 68%, 95% CL Pop-II Pop-III Pop-III, self-regulated

ma = 1 24 d

6 30

∫

zSa(z) 〈xe〉(z)− 〈xe

fix〉(z)

⎡ ⎣ ⎤ ⎦

Then evaluate optical depth

τ(z1,z2) = τ fid(z1,z2)+ maτ a(z1,z2)

∑

Source Of High Redshift Ionization: Pop-III?

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Source Of High Redshift Ionization: Pop-III?

Tanh Tanh 68%, 95% CV Pop-II Pop-III Pop-III, self-regulated

Get polarization spectrum Shades = CV assuming instantaneous ionization is the correct model* *Tricky type of question

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Make Results Useful to Everyone: PCA Fast Likelihood

0.03 0.06 0.09 0.12

τ (tanh model)

P(τ|data)

tanh tanh LPC Gaussian

L

PC(data | m) = i=1 N

∑wiK f (m − mi)

MCMC models Y

• ur Favorite

model

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

0.03 0.06 0.09 0.12

τ (tanh model)

P(τ|data)

tanh tanh LPC Gaussian

Chain multiplicities Kernel (Gaussian)

L

PC(data | m) = i=1 N

∑wiK f (m − mi)

Make Results Useful to Everyone: PCA Fast Likelihood

B-MODE from Space - Berkeley

Probing Inflation and Reionization with Large-Scale CMB Polarization

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Make Results Useful to Everyone: PCA Fast Likelihood

0.03 0.06 0.09 0.12

τ (tanh model)

P(τ|data)

tanh tanh LPC Gaussian

Good: fast (no CAMB) Bad: needed ~10x more chain points than normal convergence

L

PC(data | m) = i=1 N

∑wiK f (m − mi)

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

What Our Results Means for 21-cm, neutrinos, CMB-S4…

τ PC(z = 15,zmax) = 0.033

• CMB-S4: neutrino mass constraints with lensing
• Optical Depth is one of the highest sources of error
• 21-cm claims they can measure tau better than CV
• This claim will fail if our result is not due to systematics

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Part II - Inflationary Features Wayne Hu Chen He V . Miranda Cora Dvorkin Georges Obied

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

The Generalized Slow-Roll Approximation

I j(k) ∝ dlnsWj(ks) ′ G (lns)

∫

lnΔR

2 = I0(k)+ ln[1+ I1 2(k)]

Cora Dvorkin Wayne Hu Single kernel encompasses power spectrum observables

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Kernel Expansion: Non-Parametric Spline Basis

• C. Dvorkin W. Hu

V . Miranda G'(lns) = (1− ns)+ Bi(lns)wi

i

∑

SB: more efficient than PCAs for localized features

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

WMAP/Planck l~20 Features on Temperature Spectrum

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 ∆CTT

ℓ

/σℓ 101 102 103 ℓ −15 −10 −5 5 −2∆ ln LTT

ΛCDM TT (ℓ < 1000) SB TT

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Features Affect Inferences On The Hubble Constant

64 66 68 70 72 74 H0 [km s−1 Mpc−1] Posterior

SB TT SB TT (` < 1000) ΛCDM TT ΛCDM TT (` < 1000)

Features in inflation impact H0 predictions (especially if lmax < 1000)

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

WMAP/Planck l~20 Features On TE Spectrum

−2 −1 1 2 ∆CTE

ℓ

/σℓ 101 102 103 ℓ −15 −10 −5 5 −2∆ ln LTTEE

ΛCDM TT (ℓ < 1000) SB TT

Also impacts TE -> what about reionization?

December 4, 2017

Vinicius Miranda, Postdoctoral Researcher, University of Pennsylvania

Probing Inflation and Reionization with Large-Scale CMB Polarization

B-MODE from Space - Berkeley

Near Future: Combined Analysis (Stay Tuned)

5 10 15 20 25 30 z −0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 τ(z, zmax)

Rei Ifn+Rei

P r e l i m i n a r y ! ! ! ! !

B-MODE from Space - Berkeley - Vinicius Miranda

December 4, 2017

Conclusions

64 66 68 70 72 74 H0 [km s−1 Mpc−1] Posterior

SB TT SB TT (` < 1000) ΛCDM TT ΛCDM TT (` < 1000)

Tanh Tanh 68%, 95% CV Pop-II Pop-III Pop-III, self-regulated