2
$\begingroup$

Suppose $G$ is a finitely-generated non-elementary hyperbolic group and consider a symmetric random walk on the Cayley graph $\text{Cay}(G,S)$ with generating set $S$. Denote the set of points $B_{\ell} = \{ g: d(e,g) = \ell\}$ where $d$ is the word metric.

Let $X_n$ be the random variable denoting the position of an $n$-step symmetric random walk which starts at $e$ on the Cayley graph of a non-amenable group, I know that it has finite drift $\frac{1}{n}\lim_{n \to \infty} \mathbb{E} d(X_n,e) = v > 0$, and from (https://arxiv.org/pdf/2101.08222) $d(X_n,e)$ appears to concentrate about $v n$ if $G$ is hyperbolic.

I am wondering about properties of $\mathbb{P}(X_n = g)$ if $g \in B_{v n}$. In particular, I would not expect $\mathbb{P}(X_n = g)$ to be uniform over all points $g \in B_{v n}$ (since roughly this would require $\text{Cay}(G,S)$ to have a great number of isometries). For example, if $G$ is a free group then $\mathbb{P}(X_n = g)$ is uniform over all $g \in B_{v n}$.

Are there results in the literature where I could look to quantify how far from uniform this distribution can be if $G$ is hyperbolic? Or, are there any simple examples of $G$ where this distribution is very "far" from uniform?

$\endgroup$
2
  • 1
    $\begingroup$ I guess one way to frame this question is to compare the hitting measure of the random walk on the boundary with the Patterson-Sullivan measure (limit of uniform on balls). See arxiv.org/pdf/1501.05082 $\endgroup$
    – Corentin B
    Commented Jun 29 at 11:00
  • $\begingroup$ Thanks for the reference, seems helpful! My original intention was to quantify isoperimetric inequalities (weighted by the probability distribution of an n-step random walk $\mu_n$) of certain subsets. Namely, if I consider a spherical cap $C$ in $B_{v n}$, I was interested in $\mu_n(\partial C)/\mu_n(C)$. Having a result quantifying how close to uniform $\mu_n$ is on the surface of $B_{v n}$ would help to show that $\mu_n(\partial C)/\mu_n(C)$ is small. Are there any existing results related to this or which could be helpful? $\endgroup$
    – user8275
    Commented Jun 29 at 15:40

0

You must log in to answer this question.