A tower $\Delta_g:=\{(x,n)\in X \times \{0,1,2,\cdots\}: n < R(x)\}$

where $R:X \to \{1,2,3,\cdots\}$ is a $L^1$ function on a probability space $(X,\mu)$, $g: X \to X$ is mixing and $\gcd \{R\}=1$,

A map $f: \Delta \to \Delta$ is defined as: $f(x,n)=(x,n+1)$ if $n < R(x)-1$ and $f(x,n)=(g(x),0)$ if $n=R(x)-1$.

An invariant probability on $\mu_{\Delta}$ is $\mu_{\Delta}=(\int R d\mu)^{-1}\sum_{i\ge 0}f^i_{*}(\mu|_{R>i})$.

Is $(\Delta, f, \mu_{\Delta})$ mixing? or have a counter-example?


1 Answer 1


There are counterexamples. The easiest way to “cheat” is to let the height function be cohomologous to a constant.

As an example, let $T$ be an ergodic transformation of a space $X$. Let $A$ be a subset of $X$ such that $A$ and $T^{-1}A$ are disjoint. Now define $g(x)=1$ if $x\in A$, $g(x)=3$ if $x\in T^{-1}A$ and $2$ otherwise.

Then $T_g$ has an eigenfunction: $h(x,0)=1$ if $x\in A$, $h(x,0)=h(x,2)=-1$ and $h(x,1)=1$ if $x\in T^{-1}A$, and $h(x,0)=1$ and $h(x,1)=-1$ for other $x$’s. Then $h\circ T_g=-h$.

The way that this works is that $g$ is cohomologous to the constant function 2: the difference $g-2$ can be expressed as $j\circ T-j$ where $j(x)=1$ if $x\in A$ and 0 otherwise.

  • $\begingroup$ It is great! If the cohomology is excluded, is it possible to be mixing? My question comes from the paper by BM Gurevich in 1967 and 1969 which considered K-flow: if the base is K and the roof function satisfies some conditions like gcd=1, then the suspension flow is also K. $\endgroup$
    – jason
    Commented Nov 24, 2021 at 21:27

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.