English

On a multidimensional spherically invariant extension of the Rademacher--Gaussian comparison

Probability 2017-01-17 v3

Abstract

It is shown that \begin{equation*} \mathsf{P}(\|a_1U_1+\dots+a_nU_n\|>u)\le c\,\mathsf{P}(a\|Z_d\|>u) \end{equation*} for all real uu, where U1,,UnU_1,\dots,U_n are independent random vectors uniformly distributed on the unit sphere in Rd\mathbb{R}^d, a1,,ana_1,\dots,a_n are any real numbers, a:=(a12++an2)/da:=\sqrt{(a_1^2+\dots+a_n^2)/d}, ZdZ_d is a standard normal random vector in Rd\mathbb{R}^d, and c=2e3/9=4.46c=2e^3/9=4.46\dots. This constant factor is about 8989 times as small as the one in a recent result by Nayar and Tkocz, who proved, by a different method, a corresponding conjecture by Oleszkiewicz. As an immediate application, a corresponding upper bound on the tail probabilities for the norm of the sum of arbitrary independent spherically invariant random vectors is given.

Keywords

Cite

@article{arxiv.1603.04841,
  title  = {On a multidimensional spherically invariant extension of the Rademacher--Gaussian comparison},
  author = {Iosif Pinelis},
  journal= {arXiv preprint arXiv:1603.04841},
  year   = {2017}
}

Comments

4 pages. Version 2: fixed two typos, made a few small changes. Version 3: added introductory paragraphs, references, details, and Corollary 1