Searching for non-Gaussianity in the VSA data

Abstract

We have tested Very Small Array (VSA) observations of three regions of sky for the presence of non-Gaussianity, using high-order cumulants, Minkowski functionals, a wavelet-based test and a Bayesian joint power spectrum/non-Gaussianity analysis. We find the data from two regions to be consistent with Gaussianity. In the third region, we obtain a 96.7% detection of non-Gaussianity using the wavelet test. We perform simulations to characterise the tests, and conclude that this is consistent with expected residual point source contamination. There is therefore no evidence that this detection is of cosmological origin. Our simulations show that the tests would be sensitive to any residual point sources above the data's source subtraction level of 20 mJy. The tests are also sensitive to cosmic string networks at an rms fluctuation level of $105 \mu K$ (i.e. equivalent to the best-fit observed value). They are not sensitive to string-induced fluctuations if an equal rms of Gaussian CDM fluctuations is added, thereby reducing the fluctuations due to the strings network to $74 \mu K$ rms . We especially highlight the usefulness of non-Gaussianity testing in eliminating systematic effects from our data.