Diffusion approximations for the maximum of a perturbed random walk (Q5697196)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2214587
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Diffusion approximations for the maximum of a perturbed random walk |
scientific article; zbMATH DE number 2214587 |
Statements
Diffusion approximations for the maximum of a perturbed random walk (English)
0 references
17 October 2005
0 references
stationary sequence
0 references
0 references
0 references
0.8241240382194519
0 references
0.8118523955345154
0 references
0.8054471611976624
0 references
0.7947354316711426
0 references
0.7911021113395691
0 references
Let \(S=(S_ n)_{n\in\mathbb N_ 0}\) be a random walk, and let \(\xi=(\xi_ n)_{n\in\mathbb N_ 0}\) be a stationary sequence of random variables, then the process \(S+\xi=(S_ n+\xi_ n)_{n\in\mathbb N_ 0}\) is called a perturbation of \(S\) by \(\xi\). The paper studies the large-\(n\) behavior of the maximum, \(M_ n=\max_{k=0}^ n (S_ k+\xi_ k)\). The main novelty is the assumption of just stationarity of the perturbing sequence, which requires new ideas. The random walk \(S\) is assumed to satisfy a strong-approximation hypothesis: it can be embedded on a Brownian space and is approximated almost surely by the Brownian motion. Furthermore, a drift \(\mu\in\mathbb R\) is added to the steps of \(S\), and the distribution of \(\xi\) is allowed to depend on \(\mu\) in a certain way. With respect to \(\xi\), basically four different cases are studied. In the first case (the light-tail case with finite time horizon), it is assumed that \(\max_{k=0}^ n\xi_ k\) behaves asymptotically in distribution like \(\beta(\log n)^\gamma\) for some positive constants \(\beta\) and \(\gamma\), and the distributional limit of the process \((\mu M_{t/\mu^ 2})_{t\geq 0}\) as \(\mu\downarrow 0\) and as \(\mu\uparrow 0\) is identified. In the second case, the random variables \(\xi_ j\) are assumed to be heavy-tailed, and the limiting distribution of \(\mu M_{t/\mu^ 2}\) is found. In the third case, essentially the same assumption on \(\xi\) as in the first case is considered, but now the limiting distribution of \(| \mu| M_\infty\) as \(\mu\uparrow 0\) is examined. And finally, in the fourth case, the assumptions of the second case are in force, and the distributional limit of \(| \mu| M_\infty\) as \(\mu\uparrow 0\) is considered. Applications are outlined, and some numerical results illustrate the results.
0 references