Cross-Correlation of the Cosmic Microwave Background with the 2MASS Galaxy Survey: Signatures of Dark Energy, Hot Gas, and Point Sources

TL;DR

This work demonstrates a cross-correlation between WMAP CMB temperature maps and the 2MASS galaxy survey to extract signatures of late-time physics, namely the Integrated Sachs-Wolfe effect, thermal Sunyaev-Zeldovich signal, and microwave point sources. By modeling a three-component signal and carefully treating covariances, masking, and galaxy redshift distribution, the authors find a SZ detection at $3.1$–$3.7\sigma$, an ISW detection around $2.5\sigma$, and a microwave-source signal near $2.6\sigma$, all broadly consistent with a flat $\Lambda$CDM cosmology. The results bolster evidence for dark energy and provide constraints on hot gas content in the universe, while illustrating the importance of covariance handling and selection effects in cross-correlation analyses. The methodology and findings inform future wide-area galaxy surveys and CMB experiments aiming to sharpen ISW and SZ measurements across broader redshift ranges.

Abstract

We cross-correlate the Cosmic Microwave Background (CMB) temperature anisotropies observed by the Wilkinson Microwave Anisotropy Probe (WMAP) with the projected distribution of extended sources in the Two Micron All Sky Survey (2MASS). By modelling the theoretical expectation for this signal, we extract the signatures of dark energy (Integrated Sachs-Wolfe effect;ISW), hot gas (thermal Sunyaev-Zeldovich effect;thermal SZ), and microwave point sources in the cross-correlation. Our strongest signal is the thermal SZ, at the 3.1-3.7 σlevel, which is consistent with the theoretical prediction based on observations of X-ray clusters. We also see the ISW signal at the 2.5 σlevel, which is consistent with the expected value for the concordance LCDM cosmology, and is an independent signature of the presence of dark energy in the universe. Finally, we see the signature of microwave point sources at the 2.7 σlevel.

Figures (8)

• Figure 1: (Top panel) The histogram is the observed $K_{20}$ number-magnitude relation for galaxies in regions with $A_K < 0.05$. The solid line is the model counts inferred using data from $|b| > 30^{\circ}$ in the magnitude range $13.2 < K_{20} < 13.7$ where the extended source catalog(XSC) is most reliable. (Bottom panel) The square points gives the completeness as inferred from the difference between the observed and model counts. The solid curve is a fit to a parametric model that estimates both the incompleteness and contamination rate in a consistent manner. The dotted curve is a similar fit using data with a less stringent $A_K < 0.2$. The dashed curve is from $|b| > 30^{\circ}$, which serves roughly as the completeness upper-bound for the XSC. The vertical line at $K_{20} = 13.85$ gives a completeness at $98 \%$ for data with $A_K < 0.05$ used in our analysis.
• Figure 2: Average number of galaxies per $0.83\,\,{\rm deg}^2$ pixel (HEALpix $N_{side}$ = 64) as a function of extinction. For bright galaxies ($K_{20} < 13.5$), the galaxy density is constant up to extinction value $\sim 0.25$. For $13.5 < K_{20} < 14.0$, the density drops off at $A_K \sim 0.65$. We use only regions with $A_K < 0.05$ (dashed vertical line) for our analysis. Errors are estimated using jack-knife resampling.
• Figure 3: $dN/dz$ for the four magnitude bins used in the analysis.
• Figure 4: The auto-power for our four different magnitude bins. The solid curves show the observed auto-power multipoles with their estimated Gaussian errors (Eq.23), while the dashed curves are the projected Peacock and Dodds peacock non-linear power spectra with the best fit constant bias. The best fit Poisson noise term is subtracted out.
• Figure 5: The cross-power for our four magnitude bins. The curves are the best fit model (ISW+SZ+Point Sources) for the three bands and the points show the data. The ISW/SZ components dominate the signal for $\ell$'s below/above 20. The Point Source contribution becomes important for the lower frequency bands at the highest $\ell$'s. The shaded region shows the $1-\sigma$ error centered at the null hypothesis. Note that, while different $\ell$-bins are nearly independent, different cross-powers of bands with magnitude bins are highly correlated. As shot noise dominates the signal for our last two $\ell$-bins, for clarity, we only show the first 11 $\ell$-bins, for which the errors for the three WMAP bands are almost the same.