Asymptotic properties of conditional quantiles of the Cauchy distribution in Hilbert space (Q1291197)

The author considers a real separable Hilbert space \(H\) with inner product \(\langle.,.\rangle\) and orthonormal basis \(\{e_j\}_{j\geq 1}\). Let \(\{\lambda_j\}_{j\geq 1}\) be such that \(\lambda_j>0\), \(j\geq 1\), and \(\sum_{j\geq 1}\lambda^2_j< \infty\), and let \(\mu\) be the Cauchy distribution over \(H\) with characteristic functional \(\varphi_\mu(y)= \exp\left\{-\left(\sum_{j\geq 1}\lambda^2_j\langle y,e_j\rangle^2\right)^{1/2}\right\}\), \(y\in H\). Finally, for \(x\in H\), put \(l^2_\infty(x)= \lim_{n\to \infty} {1\over n} \sum^n_{j= 1} \langle x,e_j\rangle^2/\lambda^2_j\). The following two results reflect the reviewer's taste. Theorem: \[ \mu(\{x\in H: l^2_\infty(x)< \infty\})= 1\quad\text{and} \quad \mu(\{x\in H: l^2_\infty(x)\leq u\})= 2(1- \Phi(1/\sqrt u)),\quad u>0, \] where \(\Phi\) is the standard Gaussian distribution. Proposition: For any \(\theta\in [0,1]\), we have \[ 2\Phi(\theta\sqrt{n- 1/2})\leq 1+ B(\theta^2; 1/2,n/2)\leq 2\Phi(\theta\sqrt n/\sqrt{1- \theta^2}) \] provided that \(n\) is large enough, where \(B\) is the Beta distribution.