MathOverflow is a question and answer site for professional mathematicians. 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

Suppose that $(\Omega,\mathcal{A},\mu)$ is a $\sigma$-finite measure space of infinite measure and $T:\Omega\to\Omega$ a measure-preserving transformation with measurable inverse. Let be $\Omega_k\in \mathcal{A}$ an increasing sequence such that $\Omega_k\uparrow\Omega$ and $\mu(\Omega_k)<+\infty$ for all $k\in\mathbb{N}$.

Question 1: Given a set $A\in\mathcal{A}$, such that $\mu(A)>0$, is it true that the set $$E_k=\{\omega\in A; T^n(w)\notin A\ \forall n\in\mathbb{N}\ \text{and}\ T^{n_j}(w)\in \Omega_k \ \text{for some infinite sequence}\ (n_j(\omega)) \}$$ has zero measure ?

Question 2: If $T$ is not invertible is $\mu(E_k)=0$, in general ?

share|cite|improve this question
Your set E depends on the value of k, right? Or do you want to merely require recurrence to some $\Omega_k$, where k which may depend on $\omega$? If E depends on k, so that we have sets $E_k$, do you want $E_k$ to have zero measure for some k? For all k? – Vaughn Climenhaga Oct 30 '10 at 2:09
Hi Vaughn, yes the set E depends on k. Perhaps I should mention in the question that I defined those sets because I am trying to use them to prove that almost all points are recurrent or goes to infinity, where a point $\omega$ goes to infinity it means that for any $k\in\mathbb{N}$, there is an $n_0\equiv n_0(k)$ such that for all $n\geq n_0$ we have $T^n(\omega)\notin \Omega_k$. – Leandro Oct 30 '10 at 2:25
I don't see it in the question now, but presumably $\mu$ should be an invariant measure? – Vaughn Climenhaga Oct 30 '10 at 2:35
Yes the measure is invariant. – Leandro Oct 30 '10 at 3:03
up vote 4 down vote accepted

If I remember my infinite ergodic theory correctly, any measure-preserving transformation $T$ of a $\sigma$-finite measure space $(\Omega,\mathcal{A},\mu)$ leads to a decomposition of $\Omega$ into a dissipative part $\Omega_d$ and a conservative part $\Omega_c$. (This notation is probably non-standard.) The dissipative part is the union of the wandering sets for $T$, where a set $E$ is wandering if it is disjoint from all its preimages $T^{-n}(E)$, $n\geq 0$. The conservative part is thee complement of the dissipative part.

On the dissipative part, everything "goes to infinity" in some sense (eventually leaves $\Omega_k$, for instance), while on the conservative part, the Poincare recurrence theorem holds. (Conservative measures are exactly those $\sigma$-finite measures for which the Poincare recurrence theorem still works.) I believe this establishes the dichotomy you want.

This may even work for transformations that don't preserve $\mu$, but only preserve the collection of null sets. I'd check a reference on that, though -- details of all of this are in Aaronson's book An Introduction to Infinite Ergodic Theory.

share|cite|improve this answer
Thank you Vaughn for the answer. – Leandro Oct 30 '10 at 3:14
You're very welcome -- it took me a little while to remember this, and it's worth remembering. – Vaughn Climenhaga Oct 30 '10 at 3:16

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.