scientific article; zbMATH DE number 16048

[For the entire collection see Zbl 0713.00013.] Let $\{S_{nk},\mathcal{F}_{nk},\;k\geq 1\}$ be a zero mean, square integrable martingale, $n\geq 1$, write $X_{nk}=S_{nk}-S_{n,k-1}$. For $n\geq 1$, let $T_ n$ be a positive integer-valued stopping time w.r.t. $\{\mathcal{F}_{nk}:\;k\geq 1\}$. Let $\eta$ be a random variable with $\mathcal{P}(0<\eta<\infty)=1$; suppose $X$ is a random variable having characteristic function $\mathcal{E} \exp\{-t^ 2\eta^ 2/2\}$ and let $\mu_{2r}=\mathcal{E}| X|^{2r}$, $r>0$. Under this set up, the following results concerning convergence of moments in the martingale central limit theorem have been proved: If for some $r\geq 1$, \[ \mathcal{E}\left| \sum_{j=1}^{T_ n} \mathcal{E}(X^ 2_{nj}\mid\mathcal{F}_{n,j-1}) -\eta^ 2\right|^ r\to 0,\hbox{ and } \mathcal{E} \eta^{2r}<\infty,\tag{1} \] then the following three sets of conditions are equivalent: \[ \max_{1\leq j\leq T_ n}\mathcal{E}(X^ 2_{nj}\mid\mathcal{F}_{n,j-1})\xrightarrow{P}0,\quad\prod_{j=1}^{T_ n}\mathcal{E}(\exp(itX_{nj}))\to\exp(-t^ 2\eta^ 2/2),\quad\mathcal{E}| S_{nT_ n}|^{2r}\to\mu_{2r};\tag{2} \] \[ \max_{1\leq j\leq T_ n}\mathcal{E}(X^ 2_{nj}\mid\mathcal{F}_{n,j- 1})\xrightarrow{P}0,\quad\mathcal{E}\left|\sum_{j=1}^{T_ n} X^ 2_{nj}- \eta^ 2\right|^ r\to 0;\tag{3} \] \[ \mathcal{E}\left(\sum_{k=1}^{T_ n}| X_{nk}|^{2r}I[| X_{nk}|\geq\varepsilon]\right)\to 0, \qquad \forall\varepsilon>0.\tag{4} \] Further, under $(1)$, $S_{nT_ n}\to X$ and $\mathcal{E}| S_{nT_ n}|^{2r}\to\mu_{2r}$. Some extensions of the above assertion have also been discussed.

Recommendations

Cited in

(11)

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 Q3972695)