Zero speed for excited random walks on the event of non-transience
An excited random walk on $\mathbb{Z}^d$ moves, on its $i$-th visit to a site $x$, according to the $i$-th entry of a stack of transition probabilities attached to $x$; the stacks are independent and identically distributed over the sites. Kosygina and Zerner asked whether $X_n/n\to0$ almost surely on the event that $|X_n|$ does not tend to infinity, assuming only that the stacks are i.i.d.\ and every entry is positive. We prove that it does, in every dimension $d\ge2$ (the case $d=1$ is due to Amir, Berger and Orenshtein) and with no further restriction on the stacks. We also prove for excited random walks the form of Zerner's theorem on random walk in random environment: if, for a coordinate direction $e$, the events $X_n\cdot e\to+\infty$ and $X_n\cdot e\to-\infty$ both have probability zero, then $X_n\cdot e/n\to0$ almost surely. Together with the directional law of large numbers of Kosygina and Zerner, this gives the almost sure convergence of $X_n/n$ under the same two assumptions. The limit takes at most two values, $v$ and $-cv$ with $c\ge0$, and for every $\ell\in\mathbb{R}^d\setminus\{0\}$, $X_n\cdot\ell/n\to0$ almost surely on the event that $X_n\cdot\ell$ tends neither to $+\infty$ nor to $-\infty$. The proof replaces Zerner's use of the strong Markov property at the first visit of the walk to each hyperplane $\{x:x\cdot e=m\}$ by a sum over paths of bounded length.
Authors
- Yitzchak Shmalo (ORCID: https://orcid.org/0000-0002-2886-3049)
Institutions
- Hebrew University of Jerusalem (IL)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-13
- DOI
- https://doi.org/10.5281/zenodo.24199993
- Primary Topic
- Stochastic processes and statistical mechanics
- Type
- preprint