2
$\begingroup$

Question: Let $\varphi_t: M \to M$ be a continuous flow with no fixed points on a compact metric space $M$. If $\varphi_t$ has a periodic orbit, is there a positive lower bound on the period of all periodic orbits of $\varphi_t$? Or, can there be periodic orbits of arbitrarily small period?

For the case of a smooth flow on a smooth manifold generated by a vector field, I think that flow box theorem can produce a positive lower bound on period, but for the continuous case I'm unsure.

My attempt: A method I've tried so far is via contradiction; suppose for each $T> 0$, there is a periodic orbit with period $\leq t.$ Then, for each $n \in \mathbb N$, there is a time $t_n \in (0, 1/n)$ and a point $p_n \in M$ for which $\varphi_{t_n}(p_n) = p_n$. By identifying with a subsequence if necessary, use compactness to produce a limit point $p$ to which $p_n$ converges; however, since $t_n$ converges to zero, I can't conclude anything from this argument other than $\varphi_0(p) = p$, which is trivial. I feel like $p$ should be a fixed point, producing a contradiction, but I'm unable to prove this; any help in this direction (or an entirely different, or negative answer to the above) would be much appreciated!

$\endgroup$

1 Answer 1

1
$\begingroup$

Suppose for contradiction that for each $\epsilon > 0$, there exists a time $t \in (0, \epsilon)$ and a point $p \in M$ for which $\varphi_t(p) = p$. This implies that, for each $n \in \mathbb N$, there exists a time $t_n \in (0, \frac{1}{n})$ and a point $p_n \in M$ for which $\varphi_{t_n}(p_n) = p_n$. Since $M$ is compact, $\{p_n\}$ has a convergent subsequence $\{p_{n_k}\}$ with limit $p \in M$.

Now, fix a time $t > 0$. There exists a sequence of integers $\{a_n\} \subseteq \mathbb Z$ for which $\{a_n t_n\}$ converges to $t$ as $n \to \infty$. To see this, consider the sequence $\{a_n\} = \lfloor{\tfrac{t}{t_n}}\rfloor$, which is well-defined since $t_n \neq 0$ for each $n$. One always has $| \lfloor{\tfrac{t}{t_n}}\rfloor - \tfrac{t}{t_n}| < 1$. Now, consider: \begin{equation} |a_n t_n - t| = | \lfloor{\tfrac{t}{t_n}}\rfloor t_n - t| = |t_n| \cdot | \lfloor{\tfrac{t}{t_n}}\rfloor - \tfrac{t}{t_n}| < |t_n|. \end{equation} Passing to the limit, we conclude $a_n t_n \to t$ as $n \to \infty$. For each $n \in \mathbb N$, $\varphi_{t_n}(p_n) = p_n$ implies $\varphi_{a_n t_n}(p_n) = p_n$, since $a_n \in \mathbb Z$. Thus, we conclude that both \begin{equation} \lim_{k \to \infty} \varphi_{a_{n_k} t_{n_k}}(p_{n_k}) = \varphi_t(p) \text{ and } \lim_{k \to \infty} \varphi_{a_{n_k} t_{n_k}}(p_{n_k}) = \lim_{k \to \infty} p_{n_k} = p. \end{equation} Thus, $\varphi_t(p) = p$. Since $t > 0$ was arbitrary, this implies that $p$ is a fixed point of $\varphi_t$, a contradiction.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.