English

Two Interesting Properties of the Exponential Distribution

Statistics Theory 2015-03-04 v1 Statistics Theory

Abstract

Let X1,X2,,XnX_1, X_2,\ldots, X_n be nn independent and identically distributed random variables, here n2.n \geq 2. Let X(1),X(2),,X(n)X_{(1)}, X_{(2)}, \ldots, X_{(n)} be the order statistics of X1,X2,...,Xn.X_1, X_2,..., X_n. In this note we proved that: (I) If X1,X2,...,XnX_1, X_2,..., X_n are exponential random variables with parameter c>0,c > 0, then the "correlation coefficient" between X(k)X_{(k)} and X(k+t)X_{(k+t)} is strictly increasing in kk from 11 to m,m, and then is strictly decreasing in kk from mm to nt,n - t, here tt is a fixed integer between 11 and n3,n - 3, and m=(nt)/2m = (n - t)/2 if ntn - t is even, m=(nt+1)/2m = (n - t + 1)/2 if ntn - t is odd. We also proved that if t=n2t = n - 2, then the "correlation coefficient" between X(1)X_{(1)} and X(n1)X_{(n-1)} is greater than the "correlation coefficient" between X(2)X_{(2)} andX(n).X_{(n)}. (II) The "correlation coefficient" between X(k)X_{(k)} and X(k+t)X_{(k+t)} for the exponential random variables is always less than the "correlation coefficient" between X(k)X_{(k)} and X(k+t)X_{(k+t)} for the uniform random variables for all kk and tt such that k+tn.k + t \leq n. A combinatorial identity is also given as a bi-product. \vs

Keywords

Cite

@article{arxiv.1503.01075,
  title  = {Two Interesting Properties of the Exponential Distribution},
  author = {Robert W. Chen},
  journal= {arXiv preprint arXiv:1503.01075},
  year   = {2015}
}
R2 v1 2026-06-22T08:43:29.694Z