I encounter with two definitions for expansive continuous flows and their equivalence is unclear for me. Could anyone can explain for me please? Thanks in advance. I cite below these two definitions.

The first one in the paper "expansive one-parameter flow" of Bowen and Walters (1972): A flow $\varphi^t$ acting on a compact metric space $X$ is expansive if $\forall \epsilon, \exists \delta $ with the property that if $d(\varphi^t x, \varphi^{s(t)}y) \le \delta $ for all $t\in \mathbb{R}$ a pair of points $x,y \in X$ and a continuous function $s: \mathbb{R} \rightarrow \mathbb{R}$ with $s(0)=0$ then $y= \varphi^{t_0}x$ for some $|t_0| \le \epsilon.$

The second one in the book "Introduction to the modern theory of dynamical system" of Katok and Hasselblatt, page 125: A flow $\varphi^t$ acting on a compact metric space $X$ is expansive if there exists a positive $\delta$ such that any given continuous function $s: \mathbb{R} \rightarrow \mathbb{R}$ with $s(0)=0$ satisfying $d(\varphi^t x, \varphi^{s(t)}x) \le \delta $ for all $t\in \mathbb{R}$ then the property $d(\varphi^t x, \varphi^{s(t)}y) \le \delta $ implies $y= \varphi^{t_0}x$ for some $t_0 \in \mathbb{R}.$

  • $\begingroup$ There seem to be some quantifiers missing here: do the inequalities depending on $t$ hold for all $t$, or only for certain $t$? If only for $t$ in a certain range, then on what does the range depend? $\endgroup$ – Ian Morris Jan 9 '14 at 21:17
  • $\begingroup$ Is the bad English really from the sources that are cited here? $\endgroup$ – Michael Renardy Jan 10 '14 at 2:08
  • $\begingroup$ It might be possible that the definition are not equivalent, but agree on classical examples. $\endgroup$ – Benoît Kloeckner Jan 28 '14 at 21:53

Bowen-Walters definition implies KH. The converse is fase in general but true on certain spaces (e.g. surfaces). This is clarified in artigue's thesis where the HK type are called geometric separating flows


I think that the definition in KH book should be this:

A flow $\phi^t$ is expansive if there is $\delta>0$ such that if $x,y$ satisfy $d(\phi^t(x),\phi^{s(t)}(y))<\delta$ for every $t\in\mathbb{R}$, for some continuous map $s:\mathbb{R}\to\mathbb{R}$ with $s(0)=0$, then $y=\phi^{t_0}(x)$ for some $t_0\in\mathbb{R}$.

As far as I know, this definition appears first in a 1988's paper by O. Ruggiero.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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