English

Moderate deviations for the self-normalized random walk in random scenery

Probability 2021-03-11 v2

Abstract

Let GG be an infinite connected graph with vertex set VV. Let {Sn:nN0}\{S_n: n \in \mathbb N_0 \} be the simple random walk on GG and let {ξ(v):vV}\{ \xi(v) : v \in V \} be a collection of i.i.d. random variables which are independent of the random walk. Define the random walk in random scenery as Tn=k=0nξ(Sk)T_n = \sum_{k=0}^n \xi(S_k), and the normalization variables Vn=(k=0nξ2(Sk))1/2V_n = (\sum_{k=0}^n \xi^2(S_k))^{1/2} and Ln,2=(vVn2(v))1/2L_{n,2} = (\sum_{v \in V} \ell^2_n(v))^{1/2}. For G=ZdG= \mathbb Z^d and G=TdG = \mathbb T_d, the dd-ary tree, we provide large deviations results for the self-normalized process Tnn/(Ln,2Vn)T_n \sqrt{n}/(L_{n,2}V_n) under only finite moment assumptions on the scenery.

Keywords

Cite

@article{arxiv.2001.05736,
  title  = {Moderate deviations for the self-normalized random walk in random scenery},
  author = {Tal Peretz},
  journal= {arXiv preprint arXiv:2001.05736},
  year   = {2021}
}

Comments

Slightly changed the statement of Theorem 2.1. Restructured the presentation of the paper following referees' suggestions

R2 v1 2026-06-23T13:12:48.479Z