8
$\begingroup$

Let $T$ be an arbitrary Lebesgue measure-preserving automorphism of the unit interval $I$. Let $R_{\alpha}$ denote rotation by $\alpha$, i.e. $R_{\alpha}(x)=x+\alpha \pmod{1}$ for $x \in I$ and $\alpha \in \mathbb{R}$. Is it true that the composition $R_{\alpha} \circ T$ is ergodic for (Lebesgue) almost every $\alpha \in \mathbb{R}$?

$\endgroup$
1
  • 1
    $\begingroup$ Quite a nice question - can you say something about the motivation? $\endgroup$ Jan 27, 2014 at 2:15

1 Answer 1

9
$\begingroup$

No, for trivial reasons. Take as your $T$ the map $x\mapsto-x\mod 1$. Than the map $x\mapsto\alpha-x\mod 1$ is a reflection around $\alpha/2$, and is never ergodic.

$\endgroup$
2
  • $\begingroup$ Thanks! Good point! In arxiv.org/abs/1210.5013, the authors prove that the answer to my question is "yes" in the case that $T$ is an interval exchange map. I wonder what it is about interval exchange maps that makes it true.... $\endgroup$
    – klindsey
    Jan 27, 2014 at 4:12
  • $\begingroup$ @Klindsey: it's true that "most" IETs are uniquely ergodic. If you compose an IET with a rotation, of course you get another IET, so it's not completely unexpected that you should almost always get an ergodic transformation. $\endgroup$ Jan 27, 2014 at 5:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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