Let $G$ be a discrete group. Consider $G^{\mathbb{N}}$, the space of all sequences in $G$ equipped with product $\sigma$-algebra. Let $m:G^{\mathbb{N}} \to G^{\mathbb{N}}$ be the multiplication map defined by $$m(w)_n=w_1w_2\ldots w_n$$ The push-forward measure $\mathbb{P}_{\mu}=m_{*}\mu^{\mathbb{N}}$, where $\mu$ is a probability measure on $G$, is called the Markovian measure. In particular, $$\mathbb{P}_{\mu}\left(E=\{(w_i): w_1=g_1,w_2=w_1g_2,\ldots,w_n=w_{n-1}g_n\}\right)=\mu(g_1)\mu(g_2)\ldots\mu(g_n) $$ When equipped with $\mathbb{P}_{\mu}$, we think of $G^{\mathbb{N}}$ as the space of walks, where $w_n$ is the position of the random walk at the time $n$. We denote the space of walks by $(\Omega,\mathbb{P}_{\mu})$. A function $h: G \to \mathbb{R}$ is called $\mu$-harmonic if $$h(g)=\sum_{\gamma}\mu(\gamma)h(g\gamma)$$ We denote the space of all bounded harmonic functions by $H^{\infty}(G,\mu)$. Let $h$ be a bounded harmonic function. I want to show the following:

$(i)$ $\lim_nh(w_n)$ exists for $\mathbb{P}_{\mu}$-a.e $w \in \Omega$.

$(ii)$ Hence, for any $h \in H^{\infty}(G,\mu)$ corresponds to some $\tilde{h} \in L^{\infty}(\Omega,\mathbb{P}_{\mu})$ and $$h(e)=\int_{\Omega}\tilde{h}(w)d\mathbb{P}_{\mu}(\omega)$$

$(iii)$ Moreover, for a fixed $g$ the limit $\lim_{n \to \infty}h(gw_n)=\tilde{h}(g,w)$ exists and $$h(g)=\int_{\Omega}\tilde{h}(g,w)d\mathbb{P}_{\mu}$$

For $(i)$: Let $\mathcal{B}_n$ be the $\sigma$-algebra generated by the cylinder sets $C_{g_1,g_2,\ldots,g_n}=\{w \in \Omega: w_i=g_i \text{ for } i=1,2,\ldots,n\}$, i.e. $$\sigma_n=\langle\{C_{g_1,g_2,\ldots,g_n}: g_1,g_2,\ldots,g_n \in \Gamma\} \rangle $$

Then the $\sigma$-algebra on $\Omega$ is generated by $\cup_n\mathcal{B}_n$. We wish to show that the sequence of functions $f_n$ defined by $f_n(w)=h(w_n),f_0=h(e)$ forms a bounded martingale w.r.t$\{\mathcal{B}_n\}$. Towards that end, one checks that $$\mathbb{E}(f_{n+1}|\mathcal{B}_n)=\sum_{g \in G}\mu(g)h(w_ng)=h(w_n)=f_n(w)$$ The second equality comes from the fact that $h$ is a bounded harmonic function and the third equality comes from definition of $f_n$. I have trouble understanding how the first equality comes, I know that it comes from the way conditional expectation is defined but I am not seeing it. From the above equation, we see $\{f_n\}$ forms a bounded Martingale and hence, by Martingale Convergence Theorem, it converges a.e. I am not sure how $(ii)$ should be proven. I think $(iii)$ follows from $(ii)$, but not sure why.

Any hints for these would be greatly appreciated. Thanks for the help!!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.