scientific article; zbMATH DE number 3848315

From MaRDI portal
Publication:3317819





Some general large deviation results of Cramer's type are obtained for sums of independent random variables \(\{X_ i\}_{i=1,2,...}\) with zero mean and finite moment generating functions. The first theorem treats the case of weighted sums \(S_ n=\sum^{n}_{i=1}a_{ni}X_ i\) of i.i.d.r.v.s, where \(\{a_{ni}\}\) is a triangular array of non-negative constants. Here, under certain assumptions, \[ (*)\quad P(S_ n>x_ n(\sum^{n}_{i=1}a^ 2_{ni})^{1/2})\sim 1-\Phi(x_ n), \] \[ where\quad x_ n=o((\sum^{n}_{i=1}a^ 2_{ni})^{1/2}/(\sum^{n}_{i=1}a^ 3_{ni})^{1/3}), \] x\({}_ n\to \infty\), and \(\Phi\) denotes the standard normal d.f.. The second theorem deals with the non-i.i.d. case. Assuming that the r.v.s \(X_ i\) belong to an additively closed family, the author proves an analogous result as in (*). The third theorem is an amalgamation of the first two cases.




Cited in
(25)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3317819)