Non-Gaussian Distribution and Clustering of Hot and Cold Pixels in the WMAP Five-Year Sky

We present measurements of the clustering of hot and cold patches in the microwave background sky as measured from the W ilkinson M icrowave Anisotropy Probe (WMAP) five-year data. These measurements are compared with theoretical predictions which assume that the cosmological signal obeys Gaussian statistics. We find significant differences from the simplest Gaussian-based prediction. However, the measurements are sensitive to the fact that the noise is spatially inhomogeneous (e.g., because different parts of the sky were observed for different lengths of time). We show how to account for this spatial inhomogeneity when making predictions. Differences from the Gaussian-based expectation remain even after this more careful accounting of the noise. In particular, we note that hot and cold pixels cluster differently within the same temperature thresholds at few-degree scales. While these findings may indicate primordial non-Gaussianity, we discuss other plausible explanations for these discrepancies. In addition, we find some deviations from Gaussianity at sub-degree scales, especially in the W band, whose origin may be associated with extragalactic dust emission.