Good configurations of random walk in random scenery under two-sided conditioning
Let $(S_t)_{t\in\mathbb{Z}}$ be a nearest-neighbour random walk on $\mathbb{Z}$ with $S_0=0$ and probability $p$ of a step to the right, and let $(C_x)_{x\in\mathbb{Z}}$ be an independent field of independent fair signs. The reduced random walk in random scenery is the colour process $Y_t=C_{S_t}$. Den Hollander and Steif conjectured that, under conditioning on the colours at all nonzero times, every configuration is bad for $Y_0$ when $1/2\le p<4/5$ and every configuration is good when $4/5<p\le1$. We prove that every configuration is good when $27/28<p\le1$. More precisely, the conditional probabilities of $Y_0$ given the colours at the times $0<|t|\le m$ converge uniformly as $m\to\infty$, with an explicit modulus, to a version of the conditional probability that is continuous everywhere. For colours with $\mathbb{P}(C_x=1)=\alpha$ the same holds when $p>27/(27+4\beta^2)$, where $\beta=\min(\alpha,1-\alpha)$. The proof writes the posterior law of the walk as a gas of disjoint time intervals separated by cut times, bounds the activities of the intervals, and swaps the central part of two coupled samples between common cut times.
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.24191808
- Primary Topic
- Stochastic processes and statistical mechanics
- Type
- preprint