Asymptotic expansions for sums of nonidentically distributed Bernoulli random variables (Q1122858): Difference between revisions

Let \(\{X_{nj}\), \(1\leq j\leq n\}\) be a triangular array of independent Bernoulli distributed r.v., \[ S_ n=\sum^{n}_{j=1}X_{nj},\quad \mu_ n=E S_ n,\quad \sigma^ 2_ n=D S_ n,\quad \beta_{n3}=\sum^{n}_{j=1}E(X_{nj}-E X_{nj})^ 3,\quad F_ n(x)=P(\sigma_ n^{-1}(S_ n-\mu_ n)\leq x). \] Let \(x_{n,k}=\sigma^{-1}(k-\mu_ n+)\), \(k\in {\mathbb{Z}}\), \(n\geq 1\), and \(\tilde F_{n,k}\) be the distribution function with the following properties: \(\tilde F_ n(x_{nk})=F_ n(x_{nk})\), \(k\in {\mathbb{Z}}\); \(\tilde F_ n(x)\) is continuous in x and linear on intervals \((x_{n,k- 1},x_{n,k})\), \(k\in {\mathbb{Z}}\). Finally, let \(\Phi\) (x) and \(\phi\) (x) be the distribution function and the density function of a standard normal distribution, respectively. The following proposition is presented as the main result. Theorem: There exists a universal constant such that for all \(n\geq 1\), \[ \sup_{x}| \tilde F_ n(x)-\Phi (x)-(6\sigma^ 3_ n)^{- 1}\beta_{n3}(1-x^ 2)\phi (x)| \leq C \sigma_ n^{-2}. \] In the i.i.d. case, this theorem is applied to the theory of intermediate order statistics. The i.i.d. condition is not essential.