Asymptotic behaviors for random geometric series (Q6592157)

Consider the random geometric series \(X_\beta=\sum_{n=0}^\infty\beta^n\xi_n\) for some \(\beta\in(0,1)\), where \(\xi_0,\xi_1,\ldots\) are independent and identically distributed random variables with zero mean and unit variance. The authors begin by deriving a Berry-Esseen bound and precise moderate deviations for \(X_\beta\). They further establish functional limit theorems as \(\beta\to1\) for \(\{\tilde{X}_\beta(t):t\in(0,\infty)\}\), where\N\[\N\tilde{X}_\beta(t)=\sqrt{1-\beta^2}\sum_{k=0}^\infty\beta^{kt}\xi_k\,.\N\]\NThese include a functional central limit theorem, functional law of the iterated logarithm, and functional large-deviations principle. These are established using a bounded linear operator from the path space of an appropriate random walk into the path space of the random geometric series, allowing the authors to exploit functional limit theorems for random walks.