2
$\begingroup$

Theorem 2.3, p. 26 from W Shen, Y Yi "Almost automorphic and almost periodic dynamics in skew-product semiflows" states that if there is a semi-flow $\varphi^{t}$, $t \geq 0$, on a locally compact metric space $\mathcal{S}$ such that every point $x \in \mathcal{S}$ has a unique backward orbit then the semi-flow $\varphi^{t}$ can be extended to a flow on $\mathcal{S}$. However, the proof given in the book is not satisfactory. In the most interesting part, where one has to show the continuity of $\varphi^{-t}(x) = (\varphi^{t})^{-1}(x)$ for $t > 0$, it refers to the "inverse function theorem" from a book on topology, in which I cannot find any such statement.

I know that the continuity of the inverse will follow if $S$ is compact (this is a well-known statement) or $S$ is a topological manifold (from the invariance of domain theorem). But I don't know any statement of automatic continuity for the inverse in the case of locally compact spaces.

$\endgroup$

1 Answer 1

1
$\begingroup$

This is not a complete answer, but an attempt. There exists a continuous bijective map $f \colon \mathcal{S} \to \mathcal{S}$ of a locally compact space to itself that is not a homeomorphism. This map can be contruscted from the standard conitunous bijective map of an half-open interval $[0,1)$ to the unit circle $\mathcal{S}^{1}$ that of course cannot be a homeomorphism. We just take a locally compact space $\mathcal{S} \subset \mathbb{R}^{2}$ as a disjoint union of inifite sequence of copies of $[0,1)$, say $\mathcal{I}_{k}$, and an infinite sequence of copies of $\mathcal{S}^{1}$, say $\mathcal{S}_{k}$, where $k=1,2,\ldots$. Let $f$ map $\mathcal{I}_{k+1}$ into $\mathcal{I}_{k}$, $\mathcal{I}_{1}$ into $\mathcal{S}_{1}$ and $\mathcal{S}_{k}$ into $\mathcal{S}_{k+1}$. Clearly, $f$ is a continuous bijection, but not a homeomorphism. This shows that the reasoning in the book is, indeed, at least unsatisfactory.

However, if we try to construct from this map a continuous analog with $\varphi^{1}=f$, we will have a problem with locally compactness (as for example, in the red point on the figure), which probably occurs since the dimension of the space is increased from one to two.

enter image description here

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .