2

Is my proof correct? Any general feedback on proof writing style is also much appreciated!

1. Theorem

The infinite product of connected spaces is connected in the product topology. In particular, let $ \{X_\alpha \}_{\alpha \in A}$ be an infinite collection of connected topological spaces. Then $$ X := \prod_{\alpha \in A} X_\alpha $$ is connected, where $ (X, \tau_X) $ has the product topology.

2. Lemma

Fix $ x \in X $ and define $X_K \subseteq X$ for finite index set $K$ as

$$ X_K := \{ z \in X : \forall i \notin K, z_i = x_i \} $$

Then $x \in X_K$ and $ X_K $ is homeomorphic to $ \prod_{k \in K} X_k $.

3. Proof of Lemma

Define $ f : X_K \rightarrow \prod X_k $ as simply removing coordinates not in $K$.

This is clearly surjective. And injective: if $f(u) = 0$, then $u_k = 0$ for each $ k \in K $ (and the coordinates outside of $K$ are fixed).

Next, note that for any open $U \subset X_K$, as subspace of $X$, there exists in $V \in \tau_X$ such that $ U = V \cap X_K $. Thus, each $U_k$ is open in $X_k$ and the remaining coordinates are fixed.

Since any open $ W \subset \prod X_k $ has open $W_k$ in $X_k$, it's immediate that $f$ and $f^{-1}$ map $ U_k \leftrightarrow U_k $. It follows that f is a homeomorphism. $\square$

4. Proof of Theorem

Again fix $ x \in X $. Let $ E := C(x) $, the connected component of $x$ in $X$.

Since E is closed (we will not prove this here), it suffices to show that E is dense.

First note that $\prod X_K$ is connected for any for any finite $K$, since the finite product of connected sets is connected (not proven here). Now we use the Lemma to conclude that $X_K = f^{-1}(\prod X_k)$ is connected (connectedness is a topological property). It follows that $X_K \subseteq E $ for all finite $K$.

It remains to show that $ F:= \bigcup \{ X_K : finite~K \} $ is dense in X. This will follow from a property of the product topology. Also, note that the arbitrary union of connected sets is connected (not shown here), so $F \subseteq E$.

We will show that all points of $X$ are adherent to $X_K$ for some finite $K$ and then be done. Fix $y \in X$ and a y-neighborhood $U \in \tau_X$. From the product topology, we have $ U = \prod U_\alpha $ with $U_\alpha \ne X_\alpha $ only finitely often with $ U_\alpha$ open in $X_\alpha$.

Then let $I$ be the finite set of indices where $U_\alpha \ne X_\alpha$. Then we can choose a $z \in X_K$ with $K = I$ and $ z_k = k_k $ for each $ k \in K$.

It follows that $z \in U$ and all points of $X$ are adherent to $F \subseteq E$. $\square$

5. Related posts.

This post does not provide a complete proof but is relevant.

I think this post provides a complete proof. Thanks @Henno Brandsma.

6. Updates

Thanks to @diracdeltafunk and @Henno Brandsma for pointing out errors/confusing conventions in my original proof.

I think this version (2022/2/12 at 12:30 ET) fixes those issues, so I'll treat this as an answer unless someone raises another issue.

IsaacR24
  • 725
  • 5
  • 14
  • Your definition of $X_K$ does not make sense. For one thing, it references the symbols $X$, $x_i$, and $z_i$ which were not introduced, but even ignoring this it does not parse as a coherent mathematical definition. You could try defining $X_K$ using set-builder notation, e.g. $X_K = {x \in X : \forall i \notin K (x_i = z_i)}$ or something like that (but again, you need to explain what the $z_i$'s are). – diracdeltafunk Feb 11 '22 at 19:12
  • There seems to be another problem later -- you say "note that $X_K$ represents all finite sets $K$, not a single fixed set". This also does not make sense. Surely the meaning of $X_K$ depends on the set $K$. If you want to consider all finite subsets, you must do so explicitly. – diracdeltafunk Feb 11 '22 at 19:13
  • Also, "it follows that $X_K \subset E$" only works if you know that $x \in X_K$, but this won't always be true! – diracdeltafunk Feb 11 '22 at 19:17
  • @diracdeltafunk: I see the confusion. Use your definition of $ X _ K $ in comment 1, and change $ X_K $ to $ F = { X_K : K is finite } $. That addresses your comment 2. I think your comment 3 is incorrect as a matter of set theory. I don't need $ x \in X_K $. $ X_K $ may be a proper subset of $ E $. I've shown that $ X_K $ (contained in $ E $) is dense in $ X $, which implies that $ E $ is dense in $ X $. – IsaacR24 Feb 11 '22 at 20:29
  • Sorry let me clarify $X_K$ further (you have $x \in X$ instead of $z \in X$) . We already have a fixed $ x \in X $. $X_K := { z \in X : \forall i \notin K (x_i = z_i)}$. – IsaacR24 Feb 11 '22 at 20:38
  • my answer here is relevant.. – Henno Brandsma Feb 12 '22 at 00:12
  • You need to explicitly show that $F$ is connected too. – Henno Brandsma Feb 12 '22 at 00:14
  • I think you mean $F = \bigcup{X_K : K ~\text{is finite}}$ – diracdeltafunk Feb 12 '22 at 00:48
  • You have defined $E$ to be the connected component of $x$. By definition, this means that $E$ is the largest connected set containing $x$. So, if you want to deduce that $F \subseteq E$, you need to show that $x \in F$ and that $F$ is connected. – diracdeltafunk Feb 12 '22 at 00:51
  • Oh I see -- since your $x$ is fixed from the start, it will in fact be the case that $x \in X_K$ for all $K$. One thing that makes your proof hard to parse & understand is that it is missing crucial information, like that you are fixing an element $x \in X$ from the start, and that $X$ is meant to denote the product of an infinite collection of (presumably nonempty) connected spaces ${X_\alpha}$. – diracdeltafunk Feb 12 '22 at 00:59
  • So sorry, I said x is fixed in the Theorem Proof, but not it the Lemma Proof. I see why it created significant confusion! With that correction, is the remainder of the proof correct? I'll fix the original post tomorrow morning. – IsaacR24 Feb 12 '22 at 01:38
  • Ok my original post/proof is updated responsive to your suggestions. Let me know if it looks good now. Thanks! – IsaacR24 Feb 12 '22 at 18:44

1 Answers1

0

I'm submitting a finalized version of my original proof as a proposed answer.

1. Theorem

The infinite product of connected spaces is connected in the product topology. In particular, let $ \{X_\alpha \}_{\alpha \in A}$ be an infinite collection of connected topological spaces. Then $$ X := \prod_{\alpha \in A} X_\alpha $$ is connected, where $ (X, \tau_X) $ has the product topology.

2. Lemma

Fix $ x \in X $ and define $X_K \subseteq X$ for finite index set $K$ as

$$ X_K := \{ z \in X : \forall i \notin K, z_i = x_i \} $$

Then $x \in X_K$ and $ X_K $ is homeomorphic to $ \prod_{k \in K} X_k $.

3. Proof of Lemma

Define $ f : X_K \rightarrow \prod X_k $ as simply removing coordinates not in $K$.

This is clearly surjective. And injective: if $f(u) = 0$, then $u_k = 0$ for each $ k \in K $ (and the coordinates outside of $K$ are fixed).

Next, note that for any open $U \subset X_K$, as subspace of $X$, there exists in $V \in \tau_X$ such that $ U = V \cap X_K $. Thus, each $U_k$ is open in $X_k$ and the remaining coordinates are fixed.

Since any open $ W \subset \prod X_k $ has open $W_k$ in $X_k$, it's immediate that $f$ and $f^{-1}$ map $ U_k \leftrightarrow U_k $. It follows that f is a homeomorphism. $\square$

4. Proof of Theorem

Again fix $ x \in X $. Let $ E := C(x) $, the connected component of $x$ in $X$.

Since E is closed (we will not prove this here), it suffices to show that E is dense.

First note that $\prod X_K$ is connected for any for any finite $K$, since the finite product of connected sets is connected (not proven here). Now we use the Lemma to conclude that $X_K = f^{-1}(\prod X_k)$ is connected (connectedness is a topological property). It follows that $X_K \subseteq E $ for all finite $K$.

It remains to show that $ F:= \bigcup \{ X_K : finite~K \} $ is dense in X. This will follow from a property of the product topology. Also, note that the arbitrary union of connected sets is connected (not shown here), so $F \subseteq E$.

We will show that all points of $X$ are adherent to $X_K$ for some finite $K$ and then be done. Fix $y \in X$ and a y-neighborhood $U \in \tau_X$. From the product topology, we have $ U = \prod U_\alpha $ with $U_\alpha \ne X_\alpha $ only finitely often with $ U_\alpha$ open in $X_\alpha$.

Then let $I$ be the finite set of indices where $U_\alpha \ne X_\alpha$. Then we can choose a $z \in X_K$ with $K = I$ and $ z_k = k_k $ for each $ k \in K$.

It follows that $z \in U$ and all points of $X$ are adherent to $F \subseteq E$. $\square$

IsaacR24
  • 725
  • 5
  • 14