Cosmological results from the Wilkinson Microwave Anisotropy Probe

1
Cosmological resultsfrom the Wilkinson
MicrowaveAnisotropy Probe

• Matthew Colless
• The New Cosmology
• Physics Summer School 2003

2
(No Transcript)
3
(No Transcript)
4
The WMAP spacecraft
5
The WMAP observations

• WMAP designed to minimize systematic errors using
  differential sky measurements.
• Placed in orbit at L2, so 6 months maps the whole
  sky, so these first-year results contain two sets
  of full-sky observations.

6
Spacecraft path and orbit at L2
7
Calibrations, errors and maps

• WMAP team claim that systematic errors are well
  understood and controlled, based on multiple
  checks and detailed tests.
• Calibration is based on the Earth-velocity
  modulation of the CMB dipole claim calibration
  good to 0.5.
• Beam patterns are measured by observing Jupiter
  the uncertainties in the beam pattern affect the
  window function.

8
COBE vs WMAP

• Beamsize
• COBE 7
• WMAP 0.2
• Spectral coverage
• COBE 3 bands
• WMAP 5 bands

9
Which is the simulation?

• A simulation is compared to the real first-year
  WMAP Q-band mapwhich is which?

10
WMAP dataflow
11
Foreground corrections

• Essential to correct for Galactic emission and
  extragalactic point sources. The CMB is separated
  from the foregrounds using the spectral
  information in the five WMAP bands.
• Sky regions with bright foreground emission are
  masked.
• Low-level diffuse emission removed by forming a
  map based on a MEM linear combination of the five
  bands - but this map has complex error properties
  and is not used in the analysis.
• Cosmological parameters are derived from a map
  based on masking bright sources and subtracting
  foregrounds based on spectral templates for the
  various components (IRAS ? dust, 408MHz ?
  synchrotron radiation, H? ? free-free ionized
  gas).
• This method leaves rms foreground contaminations
  of lt7?K in the Q-band and lt3?K in the V and W
  bands for l lt15.

12
CMB foreground sources
13
WMAPs CMB anisotropy map

• Best-buy linear combination of the maps in the
  different bands, optimized to remove the
  foregrounds - but errors are complex, so not
  actually used in fitting cosmological models.

14
Limits on non-Gaussianity

• The power spectrum is only a complete statistical
  description of the CMB anisotropies only if they
  are a Gaussian field.
• Most inflationary models predict that the
  fluctuations should be Gaussian (at least at
  currently detectable levels).
• The WMAP maps are tested for non-Gaussian
  behaviour using Minkowski functionals and the
  bispectrum.
• These are used to determine the lowest-order
  non-Gaussian term in a Taylor expansion of the
  curvature perturbations.
• The non-Gaussianity is characterized in terms of
  a non-linear coupling parameter fNL (fNL 0 ?
  Gaussian)
• bispectrum gives -58 lt fNL lt 134 (95 confidence)
• Minkowski functionals give fNL lt 139 (95
  confidence)
• This is consistent with Gaussianity, but its not
  clear what values of fNL might be expected from
  non-standard models.

15
Dipole, quadrupole and multipoles

• Dipole amplitude
• COBE 3.353?0.024mK ? (l,b) (264.26?0.33,48.2
  2?0.13)
• WMAP 3.346?0.017mK ? (l,b) (263.85?0.10,48.25
  ?0.04)
• Quadrupole amplitude
• COBE Qrms 10 (7, -4) ?K
• WMAP Qrms 8 (2, -2) ?K
• Multipole amplitudes (the power spectrum)
• Computed using both a quadratic estimator and a
  maximum likelihood technique.
• The QE power spectrum is used in the analysis,
  with the ML power spectrum used only as a
  cross-check.
• First peak at l 220.1?0.8 second peak at l
  546?10.

16
The CMB power spectrum from WMAP
17
Summary of previous CMB power spectra
18
Comparison to previous CMB results

• The WMAP power spectrum is normalized 10 higher
  at large multipoles compared to previous CMB
  results. (Why?)

19
Comparison to prediction (old CMB 2dF)

• The change in normalization between the old CMB
  results and the WMAP power spectrum is
  essentially the whole difference between the old
  CMB 2dFGRS prediction and the new result.

20
CMB polarization and TE cross-correlation

• Use measurements of Stokes I parameter (Q,U to
  follow) calibrated against Taurus A.
• The temperature-polarization (TE) cross-power
  spectrum shows
• correlations on large scales (low l ) due to
  re-ionization
• correlations on small scales from adiabatic
  fluctuations

21
Re-ionization and super-horizon modes

• The re-ionization feature in the TE cross-power
  spectrum corresponds to an integrated optical
  depth ? 0.17?0.04.
• In plausible models for the re-ionization
  process, this optical depth implies
• a redshift of re-ionization of zr 20 (10, -9)
  at 95 c.l.
• an epoch of re-ionization at tr 100 - 400 Myr
  (95 c.l.)
• re-ionization suppressed acoustic peak amplitudes
  by 30
• The high value of zr is incompatible with
  significant amounts of wark dark matter - WDM
  would suppress clustering on small scales and
  delay the formation of stars and QSOs, giving a
  later epoch of re-ionization.
• The anti-correlations observed in the cross-power
  spectrum imply super-horizon-scale fluctuation
  modes, as predicted by inflationary models. (But
  is this consistent with the lack of power at low
  l in the power spectrum?)

22
Cosmological models

• A flat universe with a scale-invariant spectrum
  of adiabatic Gaussian fluctuations, with
  re-ionization, is an acceptable fit to the WMAP
  data.
• This is also an acceptable fit to the combination
  of the WMAP ACBAR CBI anisotropies, the
  2dFGRS galaxies Ly ? forest clustering data,
  the HST Key Project H0, and the SN Ia data.
• The TE correlations and the acoustic peaks imply
  the initial fluctuations were primarily adiabatic
  (the primordial ratios of dark matter/photons
  baryons/photons do not vary spatially).
• The initial fluctuations are consistent with a
  Gaussian field, as expected from most
  inflationary models.
• The WMAP data (combined with any one of the HST
  H0, the 2dFGRS ?m, or the SN Ia data) implies
  that ?tot1.02?0.02.
• The dominant constituent of the universe is dark
  energy, with ??0.73?0.04.

23
The best-fit cosmological model

• however, this simple model is not the best fit -
  can do better if a scale-dependent initial
  spectral index is included
• The best model has
• initial spectral index ns 0.93 (at k00.05
  Mpc-1, i.e. 120 Mpc)
• a variation with scale dns/dlnk -0.03 ? 0.017
  (also at k0)
• This running index implies lower amplitude
  fluctuations on the smallest scales, altering the
  dark matter profiles on these scales.
• Could this be part of the solution to the problem
  of the dark matter halo profiles in dwarf
  galaxies?

24
The standard model (WMAP 2dFGRS SN Ia
HST KP)
25
The composition of the universe
?tot 1.02 ? 0.02 ?? 0.73 ? 0.04 ?CDM
0.23 ? 0.04 ?b 0.044?0.004 ?? lt
0.015 (95)
2dFGRS old CMB ?? 0.75 ? 0.10 ?CDM
0.23 ? 0.06 ?b 0.039?0.012 ?? lt
0.035 (95)
26
WMAPs cosmic timeline

• CMB last scattering surface tdec
  379 ? 8 kyr (zdec 1089 ? 1)
• Epoch of re-ionization
  tr 100 - 400 Myr (95 c.l.)
• Age of the universe today
  t0 13.7 ? 0.2 Gyr
• Hubble constant
  H0 71 ? 4 km/s/Mpc
  (cf. HST KP H0 72 ? 7 km/s/Mpc)

27
Support for inflation and some hints

• Inflation predicts
• the universe is flat ? ?tot 1.02 ? 0.02
• Gaussian initial fluctuations ? -58 lt fNL lt 134
• nearly-flat initial spectral index ? ns 0.93
• super-horizion fluctuations ? TE correlations
• The WMAP data provide some additional hints
• the scalar spectral index is not exactly unity
  ns 0.93
• the spectral index changes with scale dns/dlnk
  -0.03
• the tensor-to-scalar ratio at k00.002 Mpc-1
  lt0.71 (95)
• the dark energy equation of state parameter w lt
  -0.78 (95)

28
A problem with the standard model?

• The standard model predicts higher values of the
  correlation function for small l - i.e. on large
  angular scales. This is best seen in the
  correlation function.

Probability of so little power at ?gt60 is 2x10-3
29
Other possible problems?

• Why is the WMAP normalization of the CMB power
  spectrum 10 higher than most previous results?
  An artefact of the way Wang et al. combined the
  previous CMB data?
• Is the lack of power at low l in the TT power
  spectrum consistent with the super-horizon-scale
  fluctuation modes inferred from the
  anti-correlations observed in the TE cross-power
  spectrum?
• Is the high redshift of re-ionization (zr 20)
  found from WMAP compatible with the observations
  of z6 QSOs which seem to suggest a more recent
  epoch of re-ionization?

• Possible resolutions
• Improvements in the WMAP maps with time (more
  data and better calibrations)
• Improvements in external datasets (e.g. from
  combining with P(k) from full 2dFGRS sample, and
  from SDSS)
• Independent analyses and more sophisticated
  modelling