Non-Gaussianity of the derived maps from the first-year WMAP data

We present non-Gaussianity testing on recently-released derived maps from the first-year WMAP data by Tegmark, de Oliveria-Costa and Hamilton. Our test is based on a phase mapping technique which has the advantage of testing non-Gaussianity at separate multipole bands. We show that their foreground-cleaned map is against the random-phase hypothesis at all 4 multipole bands centered around l = 150, 290, 400 and 500. Their Wiener-filtered map, on the other hand, is Gaussian for l < 250, and marginally Gaussian for 224 < l < 350. However, we see the evidence of non-Gaussianity for l > 350 as we detect certain degrees of phase coupling, hence against the random-phase hypothesis. Our phase mapping technique is particularly useful for testing the accuracy of component separation methods.