Let $(K,d)$ be a compact metric space and denote $\ C(K,K):=\{ f:K \to K \ ; \space f \space \text{is continuous} \}$. Endow the set $C(K,K)$ with the supremum metric $\ \rho (f,g):=\mathrm{sup} \ \{ d(f(x),g(x)) \ ; \ x \in K \}$.

I'd like to show that the group of homeomorphisms of $(K,d)$ is a $G_{\delta}$ subset of the metric space $( C(K,K),\rho)$.

I managed to show that it is enough to prove that the set $\{ f \in C(K,K) \ ; \space f \space \text{is injective} \}$ is $G_{\delta}$, but then I got stuck.

I'll appreciate any help.


Let $H$ be the set of homeomorphisms in $C(K,K)$. For $f\in H$, let $\{U_n(f)\}$ be a countable nested collection of open [in $C(K,K)$] metric balls around $f$. We will shrink the radii later. For now, let's say $r_n(f)$ is the radius of $U_n(f)$.

If we set $$ U_n =\bigcap_{f\in H} U_n(f)$$ then we want to show $H=\cap U_n$. One direction is easy, so we show the other.

Let $g\in \cap U_n$. Then we have a sequence $\{f_n\}$ with $g\in U_n(f_n)$. Let's insist that $r_n(f)<\frac{1}{n}$, so that $\{f_n\}$ is a Cauchy sequence with $f_n\rightarrow g$.

Since inversion is continuous on $H$, we can further shrink $r_n(f)$ [dependent on $f$] such that $$ \rho(f,\alpha)<2r_n(f)\implies \rho(f^{-1},\alpha^{-1}<\frac{1}{n} $$

This ensures $\{f_n^{-1}\}$ is also Cauchy, and thus $f_n^{-1}\rightarrow h\in C(K,K)$. But composition is continuous, so \begin{align*} f_n\circ f_n^{-1} &\rightarrow g\circ h\\ f_n^{-1}\circ f_n &\rightarrow h\circ g \end{align*} Since the left-hand sides are constantly the identity map, we see $h=g^{-1}$, and so $g\in H$, and we're done.

| cite | improve this answer | |

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.