0

Let $p \colon X \rightarrow Y$ be a closed continuous surjective map such that $p^{-1}\big(\{ y \} \big)$ is compact for each $y \in Y$. (Such a map is called a perfect map.) (a) Show that if $X$ is Hausdorff, then so is $Y$.

My attempt: It’s easy to check, normal $\Rightarrow$ regular $ \Rightarrow$ $T_2$ $\Rightarrow$ $T_1$. By Exercise 6, Section 31 of Munkres’ Topology, $Y$ is $T_2$. Is this proof correct?

Edit: This solution is incorrect. Following are comments of Mr.Gandalf Sauron:

I guess you are trying to ask the following :- "Since Hausdorffness is preserved under a perfect map then is normality/regularity also preserved?" . The short answer is no. A map preserving weaker properties does not imply that it will preserve stronger properties. For example the map $f:\Bbb{Z}\to\Bbb{Z}$ given by $f(x)=2x$ is a group homomorphism but is not a ring homomorphism.

So although intuitively it might make sense to conclude as such but there is no logical pathway that lets you conclude that the preservation of Hausdorff property will imply preservation of Normality . Even if you look at the converse that "Since Normality is preserved, is Hausdorffness also automatically preserved?" then again the answer is that you have no logical pathway in connecting these two. You have to prove them separately .

The problem is that you are thinking as $N \subset R\subset H$ and as $p$ maps $H \to H$ does it imply that $R \to R$ and $N \to N$. That is true but it requires proof and is not true for general functions. Just because a morphism maps the class of all Hausdorff topological spaces $H \to H$. Does not mean that a particular subclass $R$ of $H$ need be mapped to $R$ itself and nor so for $N \to N$. Again conversely it is not true that as $N$ is mapped to $N$ that $R$ will be mapped to $R$ or that $H$ will be mapped to $H$. Here $N$ and $R$ are not subsets but are subclasses of the class of all Hausdorff topological spaces . $H$ denotes Hausdorff, $N$ is normal and $R$ is regular.

user264745
  • 4,143
  • 1
    No, you only have that $X$ normal implies $Y$ normal, but you're starting with a weaker hypothesis: The $X$ is Hausdorff – Alessandro May 24 '22 at 06:52
  • @Alessandro Thank you for comment. If we were starting with stronger hypothesis, then would it be correct? E.g. assume this problem to be true, if $X$ is normal or regular, then $Y$ is normal or regular. – user264745 May 24 '22 at 06:59
  • According to Exercise 6, Section 31 yes for $X$ normal – Alessandro May 24 '22 at 07:04
  • @Alessandro that means this problem $\implies$ https://math.stackexchange.com/q/4452805/861687 and https://math.stackexchange.com/q/4450307/861687. – user264745 May 24 '22 at 07:29
  • @Alessandro Am I right? – user264745 May 24 '22 at 07:37

1 Answers1

1

It's unclear to me why would you start with a condition that is not given in the question. However yes, if $X$ was regular then $Y$ is regular . This will allow you to conclude Hausdorffness if you go by Munkres' definition of Regular which is a $T_{1}$ and $T_{3}$ space.

For a proper proof you need to go like this:-

Let $y_{1},y_{2}\in Y$ . Then $K_{1}=p^{-1}(y_{1})$ and $K_{2}=p^{-1}(y_{2})$ are compact sets and since we are in a Hausdorff space they are also closed.

Let $x\in K_{1}$ and $x'\in K_{2}$ . There exists $U_{x,x'}$ and $V_{x,x'}$ such that they are disjoint open sets containing $x$ and $x'$ respectively.

Then consider the open cover of $K_{1}$ by $\{U_{x,x'}\}_{x\in K_{1}}$ . This has a finite subcover $\{U_{x_{1},x'},...,U_{x_{n},x'}\}$ . Correspondingly we have the collection $\{V_{x_{1},x'},...,V_{x_{n},x'}\}$ . Let $\displaystyle U_{x'}=\bigcup_{j=1}^{n}U_{x_{j},x'}$ and $\displaystyle V_{x'}=\bigcap_{j=1}^{n}V_{x_{j},x'}$.

Then for each $x'\in K_{2}$ we have open sets $U_{x'}$ and $V_{x'}$ such that $x\in V_{x'}$ , $K_{1}\subset U_{x'}$ and $U_{x'}\cap V_{x'}=\phi$ .

Thus we consider the cover of $K_{2}$ by $\{V_{x'}\}_{x'\in K_{2}}$ . This has a finite subcover $V_{1},...,V_{m}$ and correspondily open sets $U_{1},...,U_{m}$. Then if we define $\displaystyle V=\bigcup_{i=1}^{m}V_{i}$ and $\displaystyle U=\bigcap_{i=1}^{m}U_{i}$ . Then $U$ and $V$ are open sets such that $K_{1}\subset U$ and $K_{2}\subset V$ and $U\cap V=\phi $.

Now we take the open sets $ W=Y\setminus p(X\setminus U)$ and $W'=Y\setminus p(X\setminus V)$. Then $y_{1}\in W$ and $y_{2}\in W'$ and $W\cap W'=\phi$ .

This proves Hausdorffness of $Y$.

  • That’s what I have done, here. “However yes, if $X$ was regular then $Y$ is regular and you can conclude Hausdorffness” statement is vague. Can you rephrase it and make it more precise? – user264745 May 24 '22 at 12:49
  • I have edited. Some books does not include $T_{1}$ in the definition of regularity. – Mr.Gandalf Sauron May 24 '22 at 12:56
  • To summarise everything. You agree that proof I showed above in this post, is wrong. Inspired by Alessandro comment, I made a claim in comment section, is that claim correct? – user264745 May 24 '22 at 13:01
  • Yes. The image of a regular space is regular and that of a normal space is normal under a perfect map. – Mr.Gandalf Sauron May 24 '22 at 13:06
  • Hang on wait a minute. It was I who answered your very same query a few days back. Why are you asking it in the comments if you already have answers posted in the site and also to your own questions. https://math.stackexchange.com/questions/4449567/chapter-11-theorem-5-2-2-of-james-dugundji-topology/4449612?noredirect=1#comment9318522_4449612 – Mr.Gandalf Sauron May 24 '22 at 13:09
  • You misinterpreted my comment. I’m not asking whether they are true statement or not, rather different way(approach) to reach same conclusions. More specifically, I asked, assume exercise7(a) section 31 to be true. Since normal $\Rightarrow$ regular $ \Rightarrow$ $T_2$ $\Rightarrow$ $T_1$. By this exercise, can we conclude https://math.stackexchange.com/q/4450307/861687 and https://math.stackexchange.com/q/4452805/861687. $p$ is perfect map. If $X$ is normal or regular, then $Y$ is normal or regular, respectively. I hope, you understand what I’m trying to say. – user264745 May 24 '22 at 13:27
  • I have answered in chat – Mr.Gandalf Sauron May 25 '22 at 11:29
  • 1
    Thank you for the answer! – user264745 May 25 '22 at 11:53
  • I’m so sorry to bother you again.I was reading example 1 of section 33, which is “The space $\Bbb{R}^2_l$ is completely regular but not normal. For $\Bbb{R}^2_l$ is product of spaces that are completely regular (in fact, normal)”. Last sentence kicked my following instinct: exercise 4 section 32 says, regular and Lindelöf space is normal. It’s a standard fact that second countable implies Lindelöf. Theorem 32.1 says, regular and second countable implies normal.Can we conclude theorem 32.1 from exercise 4 section 32 and standard fact?If no, what’s the relation between them(ex 1 & new approach). – user264745 May 26 '22 at 11:33
  • 1
    You can use the excercise to prove the theorem but not the other way around. – Mr.Gandalf Sauron May 26 '22 at 11:51
  • Ohhhh….. now I understand difference between question in this post and above question. Normal and regular implies $T_2$. So $Y$ is $T_2$, not normal or regular. Again thank you so much. – user264745 May 26 '22 at 11:57