Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Assume that $T: X \to X$ is a homeomorphism of a compact metric space to itself, and let $\mu$ be a $T$-invariant Borel probability measure on $X$. I want to show that for almost every $x \in X$ the forward and backwards orbits of $x$ have the same closures, that is $$\overline{\{T^n x : n \geq 1 \}} = \overline{\{T^{-n} x : n \geq 1\}}$$ for almost every $x$ and that the closures contain $x$.

So, I first want to start with the last statement, $x \in \overline{\{T^n x : n \geq 1 \}}$.

So, $X$ is totally bounded. Take $\epsilon_n = \dfrac1{2^n}$ so and the take one of the cover sets $B_1(x, \epsilon_1)$, we know now by Poincaré recurrence that there exists a subset $F_1$ of $B_1$ such that for $x_1 \in F_1$, $T^{n^1_k} x_1$ always lands in $F_1$. By compactness, passing to a subsequence if necessary, $T^{n^1_k} x_1$ converges to $y_1 \in \overline F_1$. $\overline F_1$ is compact and closed, hence totally bounded, so again I can find a cover with $\epsilon_2$ as the ball radius and I can pick the first ball and $x_2\in F_2$. So then I have $T^{n^2_k} x_2\to y_2$, if I continue like this (needs verification) I think I can find $x \in \bigcap \overline F_n$ (Cantor nesting lemma) and then $y_n$ should converge to $x$. Since I only did all of this countably many times the union should give me a set of full measure. (For the sequence I take the diagonal sequence)

This seems to me overly complicated and a bit fishy so I would like to know if I'm on the right track, am I missing something clear?

This is a homework assignment, so I would appreciate it if I only get hints.

share|cite|improve this question
up vote 1 down vote accepted

I think you can cleanup the argument as follows:

First, cover $X$ with finitely balls of radius $1$: $B_{1,1},\ldots ,B_{1,m_1}$. Then, by the Poincaré Recurrence Theorem, for each $B_{1,k}$, almost every $x\in B_{1,k}$ is such that $T^j(x)\in B_{1,k}$ for infinitely many positive integers $j$. Hence, for almost every $x\in X$, there is some $k$ such that $x\in B_{1,k}$ and $T^j(x)\in B_{1,k}$ for infinitely many positive integers $j$. Proceeding as you do, we see that there is an $x_1\in \overline{B_{1,k}}$ such that $x_1\in \overline{\left\{ T^n(x)|\, n\geq 1\right\}}$. Then, just as you said, we can cover $\overline{B_{1,k}}$ with finitely many balls of radius $1/2$: $B_{2,1},\ldots ,B_{2,m_2}$. Proceeding as before, we see that, for almost every $x$ for which we can find such an $x_1$, we can obtain an element $x_2$ contained in $\overline{\left\{ T^n(x)|\, n\geq 1\right\}}$ within a distance of $1/2$ from $x$. Proceeding inductively, we may construct $x_n$ contained in $\overline{\left\{ T^n(x)|\, n\geq 1\right\}}$ within a distance $1/n$ of $x$ for almost every $x$ for which we can find such an $x_{n-1}$. Thus, for almost every $x\in X$, we can construct such an $x_n$ for all $n\geq 1$. As this sequence converges to $x$, it follows that $x\in \overline{\left\{ T^n(x)|\, n\geq 1\right\}}$ for almost every $x$.

I think that's the same basic idea, but a bit cleaner.

I'll try to think about the other part of the problem if I have some time and I'll let you know if I come up with anything.

-Jonny Gleason

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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