Consider the set $N=\mathbb N\cup\{\infty\}$ together with the following topology: a subset $U$ of $N$ is open if either $\infty\notin U$ or $N\setminus U$ is finite.

(1) Describe continuous maps $\mathbb R\to N$ and $N\to \mathbb R$.

(2) Does there exist a subset of $\mathbb R$ homeomorphic to $N$?


(1) I'm not quite sure what is being asked. A continuous map is one with the property that preimages of open sets are open. We know how open sets look like in both spaces. But what exactly can I conclude about continuous maps?

(2) It looks like $N$ is compact. So the only candidates for such subsets are compact subsets of $\mathbb R$. But I guess I need to understand (1) first? If I do, I will have understand how restrictions of continuous maps look like as well, I suppose.

  • $\begingroup$ Recall that $\mathbb{R}$ is connected and the image of a connected space under a continuous map is connected. What you can say then about continuous maps $\mathbb{R} \to N$? For the other direction, can you extend a function $\mathbb{N} \to \mathbb{R}$ to a continuous map $N \to \mathbb{R}$? If no, what condition do you need? $\endgroup$ – Luca Carai Dec 3 '18 at 1:57
  • $\begingroup$ @LucaCarai So from what you said we can conclude that if $\mathbb R\to N$ is continuous, then it's image is a connected subset of $N$. I have a conjecture that only singletons are connected subspaces of $N$ (I thought how to prove this, but I'm not sure: this topology is so weird!). If this is so, then all continuous maps are constant. For the other direction, I don't know even what techniques I should use for continuous extensions. $\endgroup$ – user531587 Dec 3 '18 at 2:19
  • $\begingroup$ Go back to the definition of connected sets. Let U = $U_1 \sqcup U_2$. What does the topology you have say about U? $\endgroup$ – Joel Pereira Dec 3 '18 at 2:28
  • $\begingroup$ @JoelPereira That's how I think about it. Both $U_1, U_2$ are open. Either they both do not contain $\infty$, or they both have finite complements, or one of the does not contain $\infty$ and the other has finite complement. I don't think it says something spacial about $U$ other than $U$ is open... $\endgroup$ – user531587 Dec 3 '18 at 3:08

(1) a continuous map $f$ from $\mathbb{N} \cup \{\infty\}$ into $\mathbb{R}$ corresponds to a convergent sequence and its limit, in the sense that for any space $X$, $f: \mathbb{N} \cup \{\infty\} \to X$ is continuous iff $x_n = f(n)$ defines a sequence that converges to $f(\infty)$ in $X$. And conversely for every sequence $x_n$ in $X$ that converges to $x$, the function defined by $f(n) = x_n$ for all $n$ and $f(\infty) = x$, is continuous from $\mathbb{N} \cup \{\infty\}$ to $X$.

The other way around (from $\mathbb{R}$ to $\mathbb{N} \cup \{\infty\}$ there are only constant maps as $\mathbb{N} \cup \{\infty\}$ is totally disconnected, and $\mathbb{R}$ is connected and thus has connected image.

Any convergent sequence with limit (like $\{\frac{1}{n}: n \ge 1\} \cup \{0\}$) is homeomorphic to $\mathbb{N} \cup \{\infty\}$, as is easily checked.

| cite | improve this answer | |
  • $\begingroup$ In what sense does a continuous map correspond to a convergent sequence and its limit? What's the precise statement? $\endgroup$ – user531587 Dec 5 '18 at 2:18
  • $\begingroup$ For the homeomorphism part, I posted a separate question math.stackexchange.com/questions/3026518/… $\endgroup$ – user531587 Dec 5 '18 at 2:34
  • $\begingroup$ @user531587 I made the statement exact about sequences. $\endgroup$ – Henno Brandsma Dec 5 '18 at 2:52
  • $\begingroup$ I think you proved "and conversely" part of the first paragraph in the question referred to above (if $X$ is Hausdorff). I'm not sure how to deal with the other implication. To show that $f(n)$ converges to $f(\infty)$, we need to show that any neighborhood of $f(\infty)$ contains all $f(n)$ for $n$ large. But we know from continuity that for any nbhd of $f(\infty)$ there exists a nbhd of $\infty$ whose image lies in the nbhd of $f(\infty)$. Any nbhd of $\infty$ contains infinitely many elts of $N$, so the nbhd of $f(\infty)$ also contains infinitely many pts. Is that how the proof goes? $\endgroup$ – user531587 Dec 5 '18 at 3:55
  • $\begingroup$ @user531587 not just infinitely many; all but finitely many which is stronger and gives convergence. $\endgroup$ – Henno Brandsma Dec 5 '18 at 3:57

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.