Bayesian Approach to Gaussianity Testing and Shape Statistics

Previous abstract Next abstract

Session 71 -- Galaxy and CBR Distribution
Display presentation, Friday, January 14, 9:30-6:45, Salons I/II Room (Crystal Gateway)

[71.03] Bayesian Approach to Gaussianity Testing and Shape Statistics

Shan Luo and Ethan T. Vishniac (Dept. of Astronomy, University of Texas at Austin)

We adopted a Bayesian approach of testing the null hypothesis that the kurtosis of the counts in cells of galaxy distribution is zero to see if the primordial density distribution is Gaussian or not. We developed shape statistics for two and three dimensional data to quantify the amount of structures in large scale galaxy distributions. The significance level of the signal is calibrated in terms of both classical statistics and Bayesian statistics. We have applied our procedures to simulation samples as well as real galaxy surveys. Our results show the statistics we have developed here are sensitive to the structures like filaments and sheets. Our primary results also indicate there are more structures in the CfA survey than in the recent H+CDM model by Klypin et al.

Friday program listing