Third Year WMAP Results Dave Wilkinson. NASA/GSFC Bob Hill Gary Hinshaw Al Kogut Michele Limon Nils...

Post on 17-Dec-2015

220 views 0 download

Tags:

Transcript of Third Year WMAP Results Dave Wilkinson. NASA/GSFC Bob Hill Gary Hinshaw Al Kogut Michele Limon Nils...

Third Year WMAP Results

Dave Wilkinson

NASA/GSFCBob Hill Gary Hinshaw Al KogutMichele LimonNils OdegardJanet WeilandEd Wollack

PrincetonNorm Jarosik Lyman PageDavid Spergel.

UBCMark Halpern

ChicagoStephan MeyerHiranya Peiris

BrownGreg Tucker

UCLANed Wright

Science Team:

WMAPA partnership between NASA/GSFC and Princeton

QuickTime™ and aCinepak decompressorare needed to see this picture.

Johns HopkinsChuck Bennett (PI)

CornellRachel Bean Microsoft

Chris Barnes

CITAOlivier DoreMike Nolta

PennLicia Verde

UT AustinEiichiro Komatsu

What’s New in the Measurement?

Three times as much data, sqrt(3) smaller errors in maps: more than 50x reduction in model parameter space.

Direct measurement of CMB polarization.

Much better understanding of instrument, noise, gain, beams, and mapmaking.

One of 20

A-B-A-B B-A-B-A

Amplifiers from NRAO, M. Pospieszalski design

For temperature: measure difference in power from both sides. CMB: 30 uK rms

For polarization: measure the difference between differential temperature measurements with opposite polarity. CMB 0.3 uK rms

<ExEx> <ExEy><EyEx> <EyEy>

* ***

=0

0I/2I/2

(( )

))(+ Q/2

-Q/2U/2

U/2

Coherency matrix

Stability of instrument is critical

Physical temperature of B-side primary over three years. This is the largest change on the instrument.

Jarosik et al.

Three parameter fit to gain over three years leads to a clean separation of gain and offset drifts.

K Band, 22 GHz

Ka Band, 33 GHz

Q Band, 41 GHz

V Band, 61 GHz

W Band, 94 GHz

Compare Spectra

Cosmic variance limited to l=400.

First peak

Window function dominates difference

Reionization

Best fit model

Maps of Multipoles

Too aligned?

Too symmetric?

Summary of Temperature Maps

Data + completely new pipeline consistent with first year.

Maximum likelihood for low l (Efstathiou, Seljak et al.)

New improved power spectrum. No clear glitches, low-l less anomalous, clear second peak.

Calibration error still 0.5%

Polarization

New measurement of optical depth to the surface of last scattering.

First all sky measurement of polarized foreground emission.

Direct measurement of low-l E modes.

K Band, 22 GHz 50

Ka Band, 33 GHz

Q Band, 41 GHz

V Band, 61 GHz

CMB 6 uK

W Band, 94 GHz

Q&U Maps

Blowouts

Berkhuijsen et al.

Loops

Polarized Foreground Emission

B-field

Synchrotron emission

Starlight polarization

Dust emission

Dust grain

5 GHz Polarization & B field

Polarized Foreground Emission

B field from K band B field from model

Foreground Model•Template fits (not model just shown).•Use all available information on polarization directions.•Sync: Based on K band directions•Dust: Based on directions from starlight polarization.•Increase errors in map for subtraction.•Examine power spectrum l by l and frequency.•Examine results with different bands.•Examine the results with different models.

Ka 2.14 1.096Q 1.29 1.02V 1.05 1.02W 1.06 1.05

Band Pre-Cleaned Cleaned

4534 DOF

Table of

Raw vs. Cleaned

Maps

Galaxy masked in analysis

Mask

Use 75% of sky for cosmological analysis

High l TE

Crittenden et al.

High l EE

All direct polarization measurements to date.

Low-l TE

New noise, new mapmaking, pixel space foreground subtaction, different sky cut, different band combination.

New results consistent with original results.

New results also consistent with zero!

4 to model

Low l EE/BB “Features”

Still, though, even accounting for this, EE W-band l=5,7 is problematic. All others OK.

Low-l EE/BB

EE (solid)

BB (dash)

BB model at 60 GHz

r=0.3

Frequency space

“Spikes” from correlated polarized sync and dust.

Spectrum of Foreground Subtraction

Pre-cleaned error bars do not include 2NF term.

Recall, foreground subtraction is done on maps, not spectra.

We use QV for analysis, check with other channels.

Low-l EE/BB

EE Polarization: from reionization of first stars

BB Polarization: null check and limit on gravitational waves.

r<2.2 (95% CL) from just EE/BB

EE BB

Just Q and V bands.

OpticaL Depth

Optical Depth

Knowledge of the optical depth affects the determination of the cosmological parameters, especially ns

0.111 +/- 0.0220.100 +/- 0.0290.111 +/- 0.0210.107 +/- 0.018

0.111 +/- 0.0220.092 +/- 0.0290.101 +/- 0.0230.106 +/- 0.019

KaQVQVQVWKaQVW

Bands EE only EE +TE only

Best overall with 6 parameters

=0.088 +/- 0.031

BB r=0.3

EE

TE

TT

Approx EE/BB foreground

BB Lensing

BB inflation

New Cosmological Parameters

New analysis based primarily on WMAP alone.

Knowledge of optical depth breaks the n-tau degeneracy.

Take WMAP and project to other experiments to test for consistency.

Degeneracy

Knowledge of optical depth breaks the degeneracy

1yr WMAP

3yr WMAP

Best Fit LCDM Model

WMAP-1

WMAP-3 1.037 for 3162 DOF TT+TE+EE

Mean

= 0.92+-0.1 = 0.29+-0.07

WMAP-10.0230.1450.68…0.100.970.880.32Max L

0.02220.1280.73…0.0920.9580.770.24

Smaller error bars and better fit that year 1

WMAP-3

Max L

WMAP-3 SZ Marg0.02233 +/-0.00080.1268 +/-0.010.734 +/- 0.03…0.088 +/- 0.030.951 +/- 0.0170.744 +/- 0.0550.238 +/- 0.035

Max L, sym err

Add 2dFGRS, SDSS, CMB,SN,WL

The general trend is:

drops to 0.945-0.950 +0.015/-0/017

drops when CMB added & rises when

galaxies added A “working number” is 0.26

The scalar spectral index is 0.97+/- 0.02 Seljak et al. and 0.98+/-0.03 (Tegmark et al.) for WMAP-1 +SDSS.

What Does the Model Need?

Model needs , 8

Model needs not unity, 8

Model needs dark matter, 248

Model does not need: running, r, or massive neutrinos, le 3.

Gravitational Waves

WMAP alone, r<0.55 (95% CL)

WMAP+2dF, r<0.30 (95% CL)

WMAP+SDSS, r<0.28 (95% CL)

In all cases, n_s rises to compensate.

WMAP-1+SDSS Tegmark et alWMAP-1+SDSS+Lya Seljak et al

Similar behavior:

Inflation Parameters, No Running

Equation of State & Curvature

WMAP+CMB+2dFGRS+SDSS+SN

Interpret as amazing consistency between data sets.

Final Bits

No evidence for non-Gaussanity in any of our tests: Minkowski functionals, bispectrum, trispectrum…..

Sum of mass of light neutrinos is <0.68 eV (95% CL). Has not changed significantly.

New ILC

Now can be used for l=2,3!

However, some non-Gaussanity persists!

THANK YOU