1

I am a bit confused and was hoping the folk of MSE would help (and point out where the error is made).

Suppose $(X,\mathcal{B},\mu,R)$ and $(Y,\mathcal{C},\nu,S)$ are ergodic measure preserving dynamical systems. Let $m=\mu\otimes \nu$ and define $T:X\times Y\to X\times Y$ by $T(x,y)=(Rx,Sy)$. Note that if $A\times B\subset X\times Y$ is measurable with $T^{-1}(A\times B)=A\times B$, then $R^{-1}A=A$ and $S^{-1}B=B$ and so by ergodicity, $\mu(A),\nu(B)\in\{0,1\}$. It follows that $m(A\times B)\in\{0,1\}$ and so $m$ is ergodic.

On the other hand, this is clearly false - take $T(x,y)=x-y$ on the torus $\mathbb{R}/\mathbb{Z}\times\mathbb{R}/\mathbb{Z}$. This function is not constant, but it is invariant: it is easy to check that $T(x+\alpha, y+\alpha)=T(x,y)$ for any $\alpha$.

Thanks in advance!

1 Answers1

1

Not every measurable subset of $X\times Y$ has the form $A\times B$! Indeed, in your example, for instance, the set of $(x,y)$ such that $|x-y|<1/4$ is measurable and is a counterexample to ergodicity which is not of that form.

Eric Wofsey
  • 342,377