A local limit theorem for densities and large deviations of sums of random number variables (Q1175851)

Let \((X_n)\) be a sequence of i.i.d. random variables (r.v.) with a common distribution function (d.f.) \(F\) and let \(S_n = \sum^n_{j=1}X_ j\). Let \(\gamma_p\) be a r.v. having geometric d.f. with mass function \(P(\gamma_ p=k)=p(1-p)^{k-1}\), \(k=1,2,\ldots\), \(0<p<1\). When \(EX_1=0\), \(EX^2_1=\sigma^ 2\) and \(X_1\), \(X_2,\ldots\) and \(\gamma_ p\) are independent, it is known that \(\sigma^{-1}p^{1/2}S_{\gamma_p}\) converges in distribution to a Laplace r.v., as \(p\to 0\). Assumingg that the characteristic function of \(X_1\) is absolutely integrable, the author establishes that the density of \(\sigma^{-1}p^{1/2}S_{\gamma_ p}\) converges uniformly to the density of the limiting Laplace distribution. Also, for any set \(\{a_p,\ 0<p<1\}\) of positive numbers with \(a_p\to\infty\) as \(p\to 0\), the author obtains the lower and upper bounds (large deviation results) for \(P(\sigma^{-1}p^{1/2}S_{\gamma_p}>a_p)\).