Uniform comparison of tails of (non-symmetric) probability measures and their symmetrized counterparts with applications (Q877234)

This paper deals with tail probability comparison. Let \(M\) be a class of probability measures on a Banach space \((\mathbb{B},\|\cdot\|)\). Denoting by \(\overline\mu\) the symmetrization of \(\mu\), the authors provide two sufficient conditions for the following tail comparison ineqalities; \[ \tau\mu(\{\kappa\| x\|> t\}\leq\overline\mu(\{\| x\|> t\})\leq 2\mu(\{2\| x\|> t\}),\quad\forall t> 0, \] with some positive constants \(\tau\) and \(\kappa\), independent of \(\mu\in M\). As a corollary, they extend three classical results (the Lévy-type inequalities, the Kwapien-contraction inequality and a part of the Itô-Nisio theorem) to some class of non-symmetric independent \(\mathbb{B}\)-valued random variables, including the class of centered log-concave random variables, as well as stable or semistable ones. Several valuable examples are presented.