An intermediate Baum-Katz theorem (Q634567): Difference between revisions
From MaRDI portal
Created a new Item |
Added link to MaRDI item. |
||
links / mardi / name | links / mardi / name | ||
Revision as of 08:20, 30 January 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | An intermediate Baum-Katz theorem |
scientific article |
Statements
An intermediate Baum-Katz theorem (English)
0 references
16 August 2011
0 references
Let \(\{X_{n},n\geq 1\}\) be a sequence of i.i.d. random variables with partial sums \(S_{n},\) \(n\geq 1\). \textit{P. L. Hsu} and \textit{H. Robbins} [Proc. Natl. Acad. Sci. USA 33, 25--31 (1947; Zbl 0030.20101)], and \textit{P. Erdős} [Ann. Math. Stat. 20, 286--291 (1949; Zbl 0033.29001); ibid. 21, 138 (1950; Zbl 0035.21403)] proved that the sequence \(\{S_{n}/n,n\geq 1\}\) converges completely to zero, i.e., \(\sum_{n\geq 1}\text{P}(| S_{n}/n| \geq \varepsilon)<\infty ,\) \(\varepsilon >0,\) iff \(\operatorname{E}X_{1}^{2}<\infty \) and \(\operatorname{E}X_{1}=0.\) This classical result is the origin of a vast amount of publications dealing with such series and moment conditions, which includes a remarkable paper by \textit{L. E. Baum} and \textit{M. Katz} [Trans. Am. Math. Soc. 120, 108--123 (1965; Zbl 0142.14802)] and a recent contribution due to \textit{H. Lanzinger} [Stat. Probab. Lett. 39, No.~2, 89--95 (1998; Zbl 0910.60016)]. These two papers motivate the present work whose main result reads as follows. Theorem. Let \(\alpha >1.\) If \(\operatorname{E}[\exp \{(\log ^{+}| X_{1}|)^{\alpha }\}]<\infty \) and \(\operatorname{E}X_{1}=0,\) then \(\sum_{n\geq 1}\exp \{(\log n)^{\alpha }\}\frac{(\log n)^{\alpha -1}}{n^{2}}\text{P}(|S_{n}| \geq n\varepsilon )<\infty ,\) \(\varepsilon >1.\) Conversely, if this sum is finite for some \(\varepsilon >0,\) then \(\operatorname{E}[\exp \{(1-\delta )(\log ^{+}| X_{1}| )^{\alpha }\}]<\infty\), \(\delta >0.\) A multi-index version of this theorem is also presented.
0 references
sums of i.i.d. random variables
0 references
convergence rates
0 references
laws of large numbers
0 references
random fields
0 references