5
$\begingroup$

Suppose I have a compact metric space $(X,d)$ and let $\mathcal{X}=X^K$ be the product space. Consider the equivalence relation $\sim$ on $\mathcal{X}$ given as: for $\alpha,\beta\in \mathcal{X}$, $\alpha\sim\beta$ iff there exists a permutation $\tau$ of $\{1,\dots,K\}$ such that $\alpha_i = \beta_{\tau(i)}$ for all $i$. Consider the quotient space $\mathcal{X}/\sim$ and define a metric $\rho$ on this quotient space as $\rho([\alpha],[\beta]) = \min_{\tau} \sum_i d(\alpha_i, \beta_{\tau(i)})$.

I could show this defines a well-defined valid metric. I also know that product of compact spaces under the product metric is compact (Tychonoff's theorem) and for a compact topological space $Q$ with an equivalence relation $R$, the quotient space $Q/R$ is also compact. But are there any results showing whether $(\mathcal{X}/\sim, \rho)$ is compact under the topology induced by this metric (necessary and/or sufficient conditions if any)?

$\endgroup$
7
  • 6
    $\begingroup$ The topology on $\mathcal X/\sim$ given by the metric you describe is the same as the quotient topology. $\endgroup$ Aug 22, 2021 at 1:18
  • $\begingroup$ Hi @TomGoodwillie thank you for the response. Can you give some references/insight into why? $\endgroup$
    – Sunrit
    Aug 22, 2021 at 1:35
  • 4
    $\begingroup$ Following Tom Goodwillie's comment, I think that the following is true: let $X$ be a metric space with a $G$-action where $G$ is a finite group and the metric is $G$-invariant. Then the orbit space $X_G$ is metrizable with $d([x],[y])=\min\{d(x,gy)\vert g\in G\}$ (check that "open" balls are really open and constitute a neighborhood basis). Maybe we could generalize this to compact groups $G$. $\endgroup$
    – Z. M
    Aug 22, 2021 at 5:27
  • $\begingroup$ Another way to think about this is that the quotient is the orbit space with the Hausdorff metric $\endgroup$ Aug 22, 2021 at 8:08
  • 2
    $\begingroup$ I don't have a reference, but I think that what Z.M. asserts is a straightforward exercise. $\endgroup$ Aug 22, 2021 at 12:46

1 Answer 1

2
$\begingroup$

Here's a direct proof (although showing your metric yields the topology of $\mathcal X/\sim$ is more useful). In a metric space, compactness is equivalent to sequential compactness: every infinite sequence of points has a convergent subsequence. Let $x_n\in X^K$ correspond to an arbitrary sequence $[x_n]\in\mathcal X/\sim$. Since $X^K$ is compact, let $y_n$ be a subsequence converging to $x\in X^K$.

I claim $[y_n]$ converges to $[x]$. To see this, let $\epsilon>0$, and choose $N$ with $d(y_n(i),x(i))<\epsilon/K$ for each $1\leq i\leq K$ and $n\geq N$. Then $\rho([y_n],[x])\leq\sum_i d(y_n(i),x(i))<\epsilon$ for all $n\geq N$.


This also basically shows that the map $x\mapsto[x]$ is continuous, which is another angle, since continuous images of compact spaces are compact.

$\endgroup$

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.