Testing Gaussian random hypothesis with the cosmic microwave background temperature anisotropies in the three-year WMAP data

We test the hypothesis that the temperature of the cosmic microwave background is consistent with a Gaussian random field defined on the celestial sphere, using de-biased internal linear combination (DILC) map produced from the 3-year WMAP data. We test the phases for spherical harmonic modes with l <= 10 (which should be the cleanest) for their uniformity, randomness, and correlation with those of the foreground templates. The phases themselves are consistent with a uniform distribution, but not for l <= 5, and the differences between phases are not consistent with uniformity. For l=3 and l=6, the phases of the CMB maps cross-correlate with the foregrounds, suggestion the presence of residual contamination in the DLC map even on these large scales. We also use a one-dimensional Fourier representation to assemble a_lm into the \Delta T_l(\phi) for each l mode, and test the positions of the resulting maxima and minima for consistency with uniformity randomness on the unit circle. The results show significant departures at the 0.5% level, with the one-dimensional peaks being concentrated around \phi=180 degs. This strongly significant alignment with the Galactic meridian, together with the cross-correlation of DILC phases with the foreground maps, strongly suggests that even the lowest spherical harmonic modes in the map are significantly contaminated with foreground radiation.