Cosmic microwave background anomalies viewed via Gumbel Statistics

Abstract

We describe and discuss the application of Gumbel statistics, which model extreme events, to WMAP 5-year measurements of the cosmic microwave background. We find that temperature extrema of the CMB are well modelled by the Gumbel formalism and describe tests for Gaussianity that the approach can provide. Comparison to simulations reveals Gumbel statistics to have only weak discriminatory power for the conventional statistic: $f_{NL}<1000$, though it may probe other regimes of non-Gaussianity. Tests based on hemispheric cuts reveal interesting alignment with other reported CMB anomalies. The approach has the advantage of model independence and may find further utility with smaller scale data.