3
$\begingroup$

Let $(X, \mathcal{F}, \mu)$ be a general measure space, and let $T: X \to X$ be a measure-preserving transformation on $X$. Assume that $T$ is ergodic and satisfies the property that, for any set $A \in \mathcal{F}$ with $0 < \mu(A)$, there exists a positive integer $k$ such that $\mu(A \cap T^{-k}A) > 0$.

For any $A \in \mathcal{F}$ and $t \geq 0,~n\ge 1$, define the set $\mathcal{B}(A, t)$ as: $$ \mathcal{B}(A,t,n) := \left\{ x \in X : \sum_{k=1}^n \chi_A \circ T^k(x) \geq t \right\}, $$ where $\chi_A$ denotes the characteristic function of the set $A \in \mathcal{F}$.

For $A \in \mathcal{F}$ with $0 < \mu(A) < \infty$ and $\alpha \in (0,1)$, define the sequence $\theta_n(A, \alpha)$ as: $$ \theta_n(A, \alpha) := \sup \left\{ t \geq 0 : \mu_A\left( \mathcal{B}(A,t,n) \right) \geq \alpha \right\}, $$ where $\mu_A(B) = \frac{\mu(A \cap B)}{\mu(A)}$ for $B \in \mathcal{F}$, i.e., $\mu_A$ is the conditional probability measure induced by $A$.

For $A, B \in \mathcal{F}$ with $0 < \mu(A) < \infty$, $0 < \mu(B) < \infty$, and $0 < \alpha < \beta < 1$, I want to show that: $$ \liminf_{n \to \infty} \frac{\theta_n(A, \alpha)}{\theta_n(B, \beta)} \geq \frac{\mu(A)}{\mu(B)}. $$

First, observe that for any set $A \in \mathcal{F}$ with $0 < \mu(A) < \infty$ and $\alpha \in (0,1)$, the sequence $(\theta_n(A, \alpha))_{n=1}^{\infty}$ increases without bound as $n \to \infty$. Now, to derive a contradiction, assume that there exists a subsequence $(n_k)$ such that

$$ \lim_{k \to \infty} \frac{\theta_{n_k}(A, \alpha)}{\theta_{n_k}(B, \beta)} = \gamma,\quad \text{where }~0 \leq \gamma < \frac{\mu(A)}{\mu(B)}. $$

I seem to have encountered difficulty in proceeding further with this concept. Your guidance and insights would be greatly appreciated. Thank you for your time and assistance.

$\endgroup$
6
  • 1
    $\begingroup$ Your set $\mathcal B(A,t)$ has an $n$ in its definition. Do you want $\mathcal B(A,t)$ to be $\mathcal B_n(A,t)$? And then I'm guessing that the definition of $\theta_n$ would involve $\mathcal B_n$; not just $\mathcal B$? $\endgroup$ Commented Dec 5 at 23:57
  • $\begingroup$ Is $X$ a finite measure space or an infinite measure space? If the latter, as your post suggests, then you are misquoting the ergodic theorem. $\endgroup$ Commented Dec 6 at 0:01
  • $\begingroup$ @AnthonyQuas Here $X$ is an infinite measure space, and thank you for pointing out the dependence of $\mathcal{B}(A,t)$ on $n$. I have updated my question accordingly. $\endgroup$
    – abcdmath
    Commented Dec 6 at 6:20
  • 1
    $\begingroup$ In an infinite measure space, $\lim_{n\to\infty}\frac 1n\sum_{k=1}^n \chi_A\circ T^k(x)=0$ a.e. $\endgroup$ Commented Dec 6 at 7:44
  • $\begingroup$ @AnthonyQuas Thank you for pointing that out. I apologize for this oversight. $\endgroup$
    – abcdmath
    Commented Dec 6 at 8:44

0

You must log in to answer this question.

Browse other questions tagged .