7
$\begingroup$

Let $T:X\rightarrow X$ be a continuous map on a compact metric space $(X,d)$. Suppose that $\mu$ is ergodic with respect to $T$ and for every $x\in X$ there exists a constant $C=C(x)$ such that for every $f \in C(X), f \geq 0$, \begin{align*} \limsup_{N \rightarrow \infty} \frac{1}{N} \sum_{n=0}^{N-1} f (T^nx) \leq C \int f d\mu. \end{align*}

Show that $T$ is uniquely ergodic.

I knew the following theorem

$\textbf{Theorem}$ the following properties are equivalent.

(i) $T$ is uniquely ergodic

(ii) For every $f\in C(X)$, \begin{align*} A_N^f:=\frac{1}{N} \sum_{n=0}^{N-1} f(T^nx) \rightarrow C_f, \end{align*}
where $C_f$ is a constant independent of $x$.

I don't know how to induce the convergence of $A_N^f$ from the assumption in the problem.

Any help is appreciated...

Thank you!!

$\endgroup$
  • $\begingroup$ please see my answer below $\endgroup$ – mathworker21 Nov 3 at 23:06
0
$\begingroup$

Lemma: Let $\nu,\mu$ be mutually singular probability measures on a metric space $X$. Then $$\sup_{\substack{f \in C(X,\mathbb{R}^{\ge 0}) \\ f \not \equiv 0}} \frac{\int f d\nu}{\int f d\mu} = +\infty.$$

Proof: Take disjoint $A,B$ with $\nu(A) = 1, \mu(B) = 1$. Take $\epsilon > 0$. Since finite Borel measures on a metric space are inner regular, there are closed $A' \subseteq A$ and $B' \subseteq B$ with $\nu(A') \ge 1-\epsilon$ and $\mu(B') \ge 1-\epsilon$. By Urysohn's Lemma, there is $f \in C(X,[0,1])$ with $f \equiv 1$ on $A'$ and $f \equiv 0$ on $B'$. For this $f$, we have $\frac{\int fd\nu}{\int fd\mu} \ge \frac{1-\epsilon}{\epsilon}$.

.

Main Problem:

Suppose $T$ is not uniquely ergodic. Let $\nu$ be another ergodic measure w.r.t. $T$. By the ergodic theorem, for any $f \in C(X,\mathbb{R})$, $\frac{1}{N}\sum_{n \le N} f(T^nx) \to \int fd\nu$ for $\nu$-a.e. $x \in X$. Since $X$ is compact, $C(X,\mathbb{R})$ is separable (w.r.t. $||\cdot||_\infty$ topology), so in fact for $v$-a.e. $x \in X$, all $f \in C(X,\mathbb{R})$ satisfy $\frac{1}{N}\sum_{n \le N} f(T^nx) \to \int fd\nu$. Take one such $x$. Then, by hypothesis, there is some $C$ with $\int fd\nu \le C \int fd\mu$ for every $f \in C(X,\mathbb{R}^{\ge 0})$. This contradicts the Lemma.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.