zbMATH — the first resource for mathematics

Étude des extrêmes d’une suite stationnaire m-dépendante avec une application relative aux accroissements du processus de Wiener. (Study of the extremes of a stationary m-dependent sequence with an application to the increments of Wiener processes). (French) Zbl 0631.60031
The author investigates a stationary m-dependent sequence \(\{X_ i\}\) of random variables. In the first part, a uniform bound for \[ \mu^{- n}(u)P\{\max (X_ 1,...,X_ n)\leq n\}-1 \] is given for a certain \(\mu\) (u) and for large x with \(x<\sup \{y;P(X<y)<1\}\). This completes a previous result obtained by the author [Ann. Inst. Henri Poincaré, Nouv. Sér., Sect. B 17, 309-330 (1981; Zbl 0479.60044)]. The result is applied to improve an evaluation of the probability that the increments of the Wiener process remain under a given threshold.
In the second part the records for \(\{X_ n\}\) are considered. Let \(\{(T_ n,O_ n)\}\) be the sequence of record times and record values associated with \(\{X_ n\}\), and let \(\{(S_ n,R_ n)\}\) be the corresponding sequence associated with the sequence \(\{X_ n'\}\) of independent random variables having the same distribution as \(X_ 1\). It is shown that \(\{X_ n'\}\) can be defined so that \(S_ n=T_{n-q}\) and \(R_ n=O_{n-q}\) for an integer q and for all large n.
Reviewer: T.Mori

60F10 Large deviations
62G30 Order statistics; empirical distribution functions
60J65 Brownian motion
Full Text: Numdam EuDML