Extreme value statistics of the weak lensing convergence: 1. primordial non-Gaussianities

Abstract

The subject of this paper is the investigation of inflationary non-Gaussianities of the local type with extreme value statistics of the weak lensing convergence kappa. Specifically, we describe the influence of inflationary non-Gaussianities parameterised by fnl and gnl on the probability distribution p(kappa)dkappa of the smoothed convergence field with a Gram-Charlier series, for which we compute the cumulants kappa_n of the smoothed convergence field as a configuration space average of the weak convergence polyspectra. We derive analytical expressions for the extreme value distribution and show that they correspond very well to direct samples of extreme values from the Gram-Charlier distribution. We show how the standard Gumbel distribution for the extreme values is recovered in the limit of large sample size. We investigate the shape and position of the extreme value distribution for fnl- and gnl-type non-Gaussianity and quantify the dependence on the number of available samples, leading to the inference of non-Gaussianity parameters from observed extreme values. From the observation of single extreme values in the EUCLID weak lensing survey is is possible to place constraints on fnl and gnl of the order 10^2 and 10^5, respectively, while tnl can not be constrained in a meaningful way.