On asymptotic behavior of the convolution of distributions with regularly exponentially decreasing tails (Q2246221): Difference between revisions

The author studies the asymptotic behavior of distributions and densities of the sum of independent random variables assuming that their distributions decrease exponentially at infinity. More precisely, let $X_j, j=1,2,\dots$ be independent random variables with distribution functions $F_j(.)$, and let $\bar{F}_j(.)=1-F_j(.)$, and $Q_n(x)=P(X_1+\dots+X_n\geq x), n=1,2,\dots$. The author investigates the asymptotic behavior of the probability $Q_n(x)$ and its local version as $x\to\infty$ in the case , $\bar{F}_j(t)=e^{-\lambda_jt}l_j(t)$, where $\lambda_j$ are positive numbers and the functions $l_j(t), 1\leq j\leq n$, are regularly varying at infinity with some exponents $\rho_j$.