Improved measurements of the temperature and polarization of the CMB from QUaD

TL;DR

The QUaD re-analysis delivers improved CMB temperature and polarization measurements by removing ground contamination with ground templates, enabling full-sky power spectra with ~30% higher precision. Refined beam modeling, including sidelobes, shifts high-\ell power slightly upward, while two independent pipelines validate the results. The data reinforce the standard $\Lambda$CDM model and substantially tighten constraints on extensions, notably a small negative running of the scalar index and an upper limit on the tensor-to-scalar ratio $r$, with $r < 0.33$ (CMB alone). They also constrain parity-violating interactions and place an upper bound on the lensing $B$-mode signal. Overall, QUaD provides some of the most precise polarization measurements at $\ell \gtrsim 200$ and complements WMAP, ACBAR, and SDSS in sharpening the cosmological parameter landscape.

Abstract

We present an improved analysis of the final dataset from the QUaD experiment. Using an improved technique to remove ground contamination, we double the effective sky area and hence increase the precision of our CMB power spectrum measurements by ~30% versus that previously reported. In addition, we have improved our modeling of the instrument beams and have reduced our absolute calibration uncertainty from 5% to 3.5% in temperature. The robustness of our results is confirmed through extensive jackknife tests and by way of the agreement we find between our two fully independent analysis pipelines. For the standard 6-parameter LCDM model, the addition of QUaD data marginally improves the constraints on a number of cosmological parameters over those obtained from the WMAP experiment alone. The impact of QUaD data is significantly greater for a model extended to include either a running in the scalar spectral index, or a possible tensor component, or both. Adding both the QUaD data and the results from the ACBAR experiment, the uncertainty in the spectral index running is reduced by ~25% compared to WMAP alone, while the upper limit on the tensor-to-scalar ratio is reduced from r < 0.48 to r < 0.33 (95% c.l). This is the strongest limit on tensors to date from the CMB alone. We also use our polarization measurements to place constraints on parity violating interactions to the surface of last scattering, constraining the energy scale of Lorentz violating interactions to < 1.5 x 10^{-43} GeV (68% c.l.). Finally, we place a robust upper limit on the strength of the lensing B-mode signal. Assuming a single flat band power between l = 200 and l = 2000, we constrain the amplitude of B-modes to be < 0.57 micro-K^2 (95% c.l.).

Figures (19)

• Figure 1: Demonstration of the performance of the ground templating procedure described in the text. The figure shows maps of the Stokes $U$ polarization over the full QUaD sky area at 150 GHz, smoothed with a 5 arcmin Gaussian kernel. The map on the left is that obtained without removing ground templates and is heavily contaminated by ground pickup. Note the similarity of the contamination between the lead and trail halves of the map. The map on the right is that obtained when we include our templating procedure. Clearly the vast majority of the ground signal is successfully removed by this process. For the purposes of this illustration, in order to highlight the success of the template removal, in both cases, only the mean of each scan was removed from the TOD before mapping. However, for our cosmological analysis, we use maps which have had third-order polynomials removed from each scan (see Figure \ref{['fig:tqu_maps']}).
• Figure 2: Maps of the $T$ (top panels), $Q$ (middle panels) and $U$ (lower panels) Stokes parameters over the full QUaD sky area at 100 GHz (left) and 150 GHz (right). For display purposes only, the maps have been smoothed with a 5 arcmin Gaussian kernel. Note the difference in the color stretch used to display the temperature and polarization maps.
• Figure 3: Apodized 150 GHz QUaD polarization maps decomposed into $E$-modes (left) and $B$-modes (right). Once again, the maps have been smoothed with a 5 arcmin Gaussian kernel. The data is clearly dominated by $E$-modes. The amplitude of fluctuations present in the $B$-mode map is consistent with noise. The slight reduction in the amplitude of fluctuations towards the central RA of the field is due to the application of the apodization mask which down-weights the "seam" between the lead and trail halves of the map.
• Figure 4: QUaD beam profiles at 100 GHz (top panel) and 150 GHz (bottom panel) as measured from the QSO PKS0537-441. The radial profiles as predicted using our new beam models for both pipelines are over-plotted as the red and blue curves and show good agreement with the QSO data. Also shown for comparison is the elliptical Gaussian beam model used for our previous analysis in Paper II.
• Figure 5: Total transfer function, $F_$ (black curve) as measured in our Pipeline 1 analysis for the 150 GHz channel. Also shown are the transfer functions for timestream filtering (green), ground-template removal (blue) and beam suppression (red) in isolation. The dashed curve shows the suppression of signal due to map pixelization for a HEALPix resolution, $N_{\rm side} = 2048$. The total transfer function is the product of these four individual curves.