0

I searched the site for other proofs that may be similar to mine and couldn't find one. I was hoping someone could review my variation in particular for correctness. Thanks in advance.

Problem:

Show that if Y is compact, then the projection $\pi$$_1$: X $\times$ Y -> X is a closed map.

Proof:

Let $\pi$$_1$: X $\times$ Y -> X be the projection function.
Let Y be compact. Let A be a closed set of X $\times$ Y.

Then A can be written as: X $\times$ Y - C where C is an open set in X $\times$ Y.

Note: C can be written as the union of basis elements. We can then just consider some open set of the form U $\times$ V for U open in X and V open in Y.

Then consider the infinite union of open sets: $\bigcup$$_{i=0}^\infty$U$_i$ $\times$ V$_i$.

Then two cases arise here:

case 1: $\bigcup$$_{\in I}V_i$ does not cover Y.

case 2: $\bigcup$$_{\in I}V_i$ does form an open covering of Y.

In case 1 we are left with: $\pi$$_1$(X $\times$ Y - $\bigcup$$_{i \in I}U_i$ $\times$ V$_i$) = X since the RHS of the difference does not contain all the x value combinations with Y for any particular x. Hence since X is closed in X, $\pi$$_1$ is a closed map in this case.

In case 2 we can take a finite covering Y of Y by compactness = $\bigcup$$_{i=0}^n$V$_i$. If we take the corresponding intersection:

D = $\bigcap$$_{i=0}^n$U$_i$ where each i corresponds to a V$_i$ that we chosen in our finite cover. Then in D we have the points in X that are removed completely in the calculation: X $\times$ Y - C. Note: removed completely refers to the fact that D represents the points that are of the form ($x_1$,Y) for $x_1$ $\in$ X and thus are completely removed from the operation X $\times$ Y - C. So those particular x coordinates from D will not be present in any tuple in X $\times$ Y - C.

We can repeat this procedure for EVERY open covering of Y that the set V$_i$ may hold and then compute the intersection of the associated U$_i$'s. Each result is an open set of points to be removed from X in the calculation X $\times$ Y - C. Since each $\bigcap$$_{i=0}^n$U$_i$ (the intersection result) is open, then the union of all the intersection results is also open. Call this final open set B.

Finally, we have that $\pi$$_1$(X $\times$ Y - $\bigcup$$_{i \in I}U_i$ $\times$ V$_i$) = X - B since we are only interested in the x values of the resulting set = (X $\times$ Y - $\bigcup$$_{i \in I}U_i$ $\times$ V$_i$) we must only consider x values on the RHS that attain all the points in ($x_1$, Y) for $x_1$ $\in$ X. For any $x_2$ that does not have all possible combinations with Y as $x_1$ does, the compliment operation will not fully remove that $x_2$ coordinate from the space X $\times$ Y. Thus the $x_2$ coordinate will thus still be in the projection result. Hence to compute the projection result we have X - B since only B values are fully removed (i.e represented in a form of ($x_1$,Y).

So in case two we have a closed map as well.

The case where we have a finite union of open sets in X $\times$ Y is just an easier case of the above. Hence $\pi$$_1$ is a closed map.

H_1317
  • 1,085
  • 2
    \times is the product symbol in mathjax – Angina Seng Mar 02 '19 at 05:37
  • If the main point of your question is to ask for verification and critique of your own proof (as opposed to asking for any proof of this claim), you should use the ([tag:proof-verification]) tag. – Martin Sleziak Apr 25 '19 at 15:09

1 Answers1

1

I would not accept this as a proof: it's way too sloppy. Some critique points:

  • You suggest that an infinite union is countable (it could well be much bigger).

  • You don't give a proof for the claim the projection of the set equals $X$ when the $V_i$ don't cover $Y$. Just some handwaving.

  • You don't remove "isolated points" in the computation. Introduce notation. Explicitly use the compactness of $Y$. Why do you suddenly talk about all covers of $Y$ that contain some $V_i$?

  • Define the set $B$ properly, and prove the equality (if it holds at all).

There are much better written proofs available on this site, e.g. in this thread and the simplest ones work from the assumption that $\pi[A]$ is not closed and use one cover of $Y$ and its finite subcover to get a contradiction.

Henno Brandsma
  • 242,131
  • I'll admit its a bit sloppy, however I still believe its correct. On bullet number 1, I think taking out i = 0 would help imply not necessarily countable? On bullet 2, its obvious to me, though I was the one who just did the proof -- I suppose I could elaborate, thanks. On bullet 3, I do use compactness fairly clearly. I'm not using the definition of isolated points in the topological sense, so thats just bad wording on my part. I think u got that though. Considering all covers of Y that V_i can produce is necessary to me, do u think its not? – H_1317 Mar 02 '19 at 06:23
  • On bullet 4, are u saying the equality that B is open? or that the projection equals X - B ? – H_1317 Mar 02 '19 at 06:24
  • @H_1317 That it equals $X-B$. – Henno Brandsma Mar 02 '19 at 06:25
  • @H_1317 As a general point (for this site): fix your MathJax/LaTeX. – Henno Brandsma Mar 02 '19 at 06:27
  • $\bigcup_{i \in I} (U_i \times V_i)$ could be used for a general union, e.g. – Henno Brandsma Mar 02 '19 at 06:29
  • @H_1317 You don't really define $B$. Write it as a (set theoretic) formula. – Henno Brandsma Mar 02 '19 at 06:33
  • I made some edits. Included a bit more explanation on why the projection equals X - B – H_1317 Mar 02 '19 at 07:04
  • @H_1317 you still haven’t defined $B$ properly not shown a complete proof ( two inclusions eg) why the projection would equal $X-B$. – Henno Brandsma Mar 02 '19 at 07:36
  • do you not believe the projection equals X - B? – H_1317 Mar 02 '19 at 08:53
  • @H_1317 I don't know what $B$ is without a proper definition. So it's unverifiable. – Henno Brandsma Mar 02 '19 at 09:04