Sharp estimates for the median of the \(\Gamma\)(\(n\)+1,1) distribution (Q2483853)

The aim of this note is to provide new proofs for two conjectures of \textit{J. Chen} and \textit{H. Rubin} [Stat. Probab. Lett. 4, 281--283 (1986; Zbl 0596.62015)] referring to the median of the Poisson distribution. This result involves sharp upper and lower bounds for the median of the \(\Gamma(n + 1, 1)\) distribution. The median of a random variable \(X\) is defined as \(\text{Med}(X) = \inf\{ z \epsilon N \mid P(X \leq z) \geq 1/2\}\). Let \(\lambda_n\) denote the median of the \(\Gamma(n + 1, 1)\) distribution, \(n \in N\), and let \((N_\lambda, \lambda \geq 0)\) be a standard Poisson process. The above quoted paper considered the following interesting conjectures on the median of the Poisson distribution with mean \(\lambda\): Conjecture 1: \(-\log 2 \geq \text{Med}(N_\lambda) - \lambda < 1/3, \lambda > 0;\) Conjecture 2: \((\lambda_n - n)_{n \in N}\) is a decreasing sequence. \textit{K. P. Choi} [Proc. Am. Math. Soc. 121, 245--251 (1994; Zbl 0803.62007)] has shown the following inequalities: (Choi-94): \(n + 2/3 < \lambda_ n \leq \min\{ n + \log2, n + 2/3 + 1/(2n + 2) \}\), \(n \in N,\) which proves Conjecture 1, while Conjecture 2 was proved by \textit{S. E. Alm} [Bernoulli 9, 351--371 (2003; Zbl 1015.62007)]. The precise purpose and the main result of the present paper is to show that Conjectures 1 and 2 above receive an immediate proof by sharpening the inequalities Choi-94. As a by-product of this outcome, the inequality in Conjecture 1 is also improved. The proof of this theorem is using a differential calculus for not necessarily smooth functions of the standard Poisson process and the central limit theorem.