Fluctuation behavior of Poisson triangular arrays (Q799296)

The fluctuation behavior of the sequence of row sums \(S_ n=\sum^{n}_{k=1}X_{nk}\) of a triangular array of Bernoulli random variables (r.v.) \(\{X_{nk}\), \(k=1,...,n\), \(n=1,2,...\}\) was investigated by \textit{G. Baxter} [Proc. Am. Math. Soc. 6, 177-181, (1955; Zbl 0066.376)] by analogues of the law of the iterated logarithm of the form \(P(\lim \sup_{n\to\infty }(S_ n-ES_ n)/\phi_ n)=1\) for a completely independent array (I.A.) and for an array of independent columns with \(X_{kk}\geq X_{k+1,k}\geq...(IC.DC.A.)\) resulting in \(\phi_ n=\log n/\log_ 2n\) and \(\phi_ n^*=\log_ 2n/\log_ 3n\), respectively. The present paper states two analogous theorems for triangular arrays of Poisson r.v. and applies the results to the study of the fluctuation behavior of a sequence \(\{Y_ n\}\) of Poisson r.v. being either i.i.d. or being only identically distributed but having a dependency expressed by the IC.DC. property, which would then cause the slower increasing sequence \(\phi_ n^*\).