Generalized one-sided laws of the iterated logarithm for random variables barely with or without finite mean. (English) Zbl 0712.60029
Let X be a random variable and let \(\mu (x)=\int^{x}_{0}P[| X| >t]dt\) provided E \(| X| =\infty\) and \({\tilde \mu}(x)= \int^{\infty}_{x}P[| X| >t]dt\) when E \(| X| <\infty\). Assume that \(\mu\) and \({\tilde \mu}\) are slowly varying at infinity and denote by \(c_ x\) and \(\tilde c_ x\) the inverse functions of \(\mu\) and \({\tilde \mu}\), resp. Next, put \(S_ n=\sum^{n}_{k=1}k^{\alpha}X_ k\), \(b_ n=n^{\alpha}c_ n\) and \(\tilde b_ n=n^{\alpha}\tilde c_ n\). The following two main results are proved:
1. Let \(\{X,X_ n,n\leq 1\}\) be i.i.d. r.v.’s with E \(| X| =\infty\) and \(E=(X^-/\mu (X^-))<\infty\). If \(\mu\) (x)\(\sim \mu (x \log \log x)\), then for each \(\alpha >-1:\) \[ \liminf_{n\to \infty}S_ n/b_ n=(\alpha +1)^{-1}\text{ and } \limsup_{n\to \infty}S_ n/b_ n=\infty \quad a.s. \] 2. Let \(\{X,X_ n,n\geq 1\}\) be integrable, unbounded i.i.d. mean zero r.v.’s with \(E(X^-/{\tilde \mu}(X^- ))<\infty\). If \({\tilde \mu}\)(x)\(\sim {\tilde \mu}(x \log \log x)\) and \({\tilde \mu}(\tilde b_ n)=0({\tilde \mu}(\tilde c_ n))\) for \(\alpha\in (-1,-1/2)\), then \[ \liminf_{n\to \infty}S_ n/\tilde b_ n=-(\alpha +1)^{-1}\text{ and } \sup_{n\to \infty}S_ n/\tilde b_ n=\infty \quad a.s. \] whenever \(\alpha >-1\). Moreover, it is shown that if \(\{X,X_ n,n\geq 1\}\) are i.i.d. r.v.’s with E \(| X|^ p<\infty\) for all \(p<1\), then \(S_ n\) converges a.s. for each fixed \(\alpha <-1\), so that \(S_ n/b_ n\to 0\) with probability 1 whenver \(| b_ n| \to \infty\).
1. Let \(\{X,X_ n,n\leq 1\}\) be i.i.d. r.v.’s with E \(| X| =\infty\) and \(E=(X^-/\mu (X^-))<\infty\). If \(\mu\) (x)\(\sim \mu (x \log \log x)\), then for each \(\alpha >-1:\) \[ \liminf_{n\to \infty}S_ n/b_ n=(\alpha +1)^{-1}\text{ and } \limsup_{n\to \infty}S_ n/b_ n=\infty \quad a.s. \] 2. Let \(\{X,X_ n,n\geq 1\}\) be integrable, unbounded i.i.d. mean zero r.v.’s with \(E(X^-/{\tilde \mu}(X^- ))<\infty\). If \({\tilde \mu}\)(x)\(\sim {\tilde \mu}(x \log \log x)\) and \({\tilde \mu}(\tilde b_ n)=0({\tilde \mu}(\tilde c_ n))\) for \(\alpha\in (-1,-1/2)\), then \[ \liminf_{n\to \infty}S_ n/\tilde b_ n=-(\alpha +1)^{-1}\text{ and } \sup_{n\to \infty}S_ n/\tilde b_ n=\infty \quad a.s. \] whenever \(\alpha >-1\). Moreover, it is shown that if \(\{X,X_ n,n\geq 1\}\) are i.i.d. r.v.’s with E \(| X|^ p<\infty\) for all \(p<1\), then \(S_ n\) converges a.s. for each fixed \(\alpha <-1\), so that \(S_ n/b_ n\to 0\) with probability 1 whenver \(| b_ n| \to \infty\).
Reviewer: A.M.Zapala
MSC:
60F15 | Strong limit theorems |
60F05 | Central limit and other weak theorems |
60G50 | Sums of independent random variables; random walks |
References:
[1] | Adler, A., and Rosalsky, A. (1989). On the Chow-Robbins ?fair? games problem.Bull. Inst. Math. Acad. Sinica. 17, 211-227. · Zbl 0696.60030 |
[2] | Adler, A. (1989). Exact sequences for sums of independent random variables.Proceedings of the International Conference on Almost Everywhere Convergence in Probability and Ergodic Theory, Academic Press, Boston, 11-29. · Zbl 0686.60020 |
[3] | Klass, M., and Teicher, H. (1977). Iterated logarithmic laws for asymmetric random variables barely with or without finite mean.Ann. Prob. 5, 861-874. · Zbl 0372.60042 · doi:10.1214/aop/1176995656 |
[4] | Adler, A., and Rosalsky, A. (1990). On the weak law of large numbers for weighted sums of i.i.d. random variables.International J. Math. and Math. Sci. 13, (to appear). · Zbl 0727.60022 |
[5] | Chow, Y. S., and Teicher, H. (1988).Probability Theory: Independence, Interchangeability, Martingales, Springer-Verlag, New York. · Zbl 0652.60001 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.