A0161
Title: Quantitative limit theorems and bootstrap approximations for empirical projection
Authors: Moritz Jirak - University of Vienna (Austria) [presenting]
Abstract: Given a sequence of random variables in some Hilbert space, the problem of finding distributional approximations for empirical projections is considered based on the empirical covariance operator and its spectral decomposition. Previously, this problem has been studied to a large degree only in a Gaussian setup. Novel quantitative limit theorems and bootstrap approximations are presented subject only to rather mild moment conditions. In many cases, these results approve upon even their Gaussian counterparts.
