Non-Gaussianity in Cosmology

Wednesday, April 28, 2010

Non-Gaussian estimation in the CMB
All theories of structure formation in the Universe make predictions of a statistical nature. They predict the properties of the probability distribution function (PDF) of cosmological perturbations. For Inflation models the statistics are very close to Gaussian, and most of the analysis of the CMB data relies heavily on this assumption. Although this simplifies the information extraction from the CMB map, there is no well established argument for considering only pure Gaussian maps. It is therefore of extreme importance to test this assumption by analyzing existing CMB datasets. I’ve developed estimators for the bispectrum based on fast Fourier and Maximum Likelihood methods and applied them to the MAXIMA-1 dataset. This procedures can also be easily extended to larger datasets such as the upcoming Planck thus providing tighter constraints on theories of structure formation. The statistical methods developed for the CMB can also be applied to large scale structure data, where the analysis of non-Gaussianity, specially in the non-linear regime, could provide valuable information about structure evolution, such as the effects of biasing.

Non-Gaussian theories of Structure Formation
I am also interested in research concerning theoretical predictions for the higher order moments of the PDF of cosmological perturbations, from models of structure formation. Although Inflation predicts Gaussian fluctuations to a very good approximation, there is the possibility that other theories (or variations to Inflation) may lead to non-Gaussianity. I’ve been investigating how well future CMB experiments like Planck can constrain these other models through their non-Gaussian properties. Non-Gaussianity can also affect the way objects collapse in our Universe which may have interesting observational consequences.