Estimates for order statistics in terms of quantiles (Q2313799)

The authors give sharp estimates for order statistics of independent, not necessarily identically distributed random variables in terms of quantiles. Let \(X_{1},\dots,X_{n}\) be independent nonnegative random variables with distribution function \(F_{1},\dots,F_{n}\). Given \(1\leq k\leq n\), denote by \(k\!-\min_{1\leq i \leq n} X_{i}\) the \(k\)th smallest order statistics of \(X_{1},\dots,X_{n} \). Let \(F=\frac{1}{n}\sum_{i=1}^{n} F_{i}\) be the averaged distribution. Suppose that there exists a constant \(K>1\) such that \[ \frac{F(Kt)}{1-F(Kt)}\geq \frac{2F(t)}{1-F(t)} \] for all \(t>0\). The authors give examples when this condition is satisfied. The main result of the paper is the following: \[ K^{-10} q_{F} \left( \frac{k-1/2}{n}\right) \le \mathrm{Med} \left(k-\min_{1\leq i \leq n} X_{i}\right) \leq K^{13} q_{F} \left( \frac{k-1/2}{n}\right), \] where \(q_{F}(t)\) is the quantile of order \(t\) with respect to \(F\). (The symbol \(\mathrm{Med}(Y)\) denotes the median of \(Y\).)