4
$\begingroup$

Suppose $X$ is compact and Hausdorff and that $f:X \to Y$ is continuous, closed, and surjective. How can I show that $Y$ is Hausdorff?

$\endgroup$
  • $\begingroup$ Do you know what you need to prove in order to show that? Any ideas or tries? $\endgroup$ – Alejandro Nasif Salum Jan 8 '18 at 2:04
  • $\begingroup$ Try reading here. $\endgroup$ – Michael Lee Jan 8 '18 at 4:01
1
$\begingroup$

Since $X$ is compact Hausdorff, $X$ it is normal. Next, since $f$ is closed, the single-element subsets of $Y$ are closed (as $\{y\} = f''\{x\}$ for some $x \in X$.) It follows that for each pair of distinct $y_0, y_1 \in Y$, the sets $X_0 = f^{-1}(y_0)$ and $X_1 = f^{-1}(y_1)$ are disjoint closed subsets of $X$. Noting that $X$ is normal, you can separate $X_0$ and $X_1$ using open-sets with disjoint closures. The remaining bit is to turn the open-sets separating $X_0$ and $X_1$ into open subsets of $Y$ separating $y_0$ and $y_1$.

$\endgroup$
  • $\begingroup$ "Turning $X_0$ and $X_1$ into open sets.... " seems problematic. Can you finish this, or give a hint? $\endgroup$ – DanielWainfleet Jan 9 '18 at 15:33
  • $\begingroup$ Given any separation of $X_0$ and $X_1$, by disjoint open sets, $U_0$ and $U_1$. The image under $f$ (a closed map) of the complement of $U_k$, namely $C_k = f''(X\backslash U_k)$ is a closed subset of Y with $y_{1-k}$ \in C_k$. and $y_k\not\in C_k$. Hence $\endgroup$ – Not Mike Jan 9 '18 at 15:53
  • $\begingroup$ .. letting $V_k = Y \backslash C_k$ be the open complement of $C_k$, you have $y_k \in V_k$ and $y_{1-k} \in V_{1-k}$. Moreover, $V_{k} \cap V_{1-k} = Y \backslash (C_k \cup C_{1-k}) =\emptyset$ $\endgroup$ – Not Mike Jan 9 '18 at 16:04
1
$\begingroup$

Let $y_1 \neq y_2$ be distinct points of $Y$. By surjectivity we have $x_1 ,x_2 \in X$ such that $f(x_1) = y_1 , f(x_2) = y_2$. In particular, $f[\{x_i\}] = y_i$ for both $i$ and as in $X$ singletons are closed (Hausdorff implies $T_1$) and $f$ is a closed map (first time we use it), we have that both $\{y_i\}$ are closed in $Y$.

Now, define $F_i = f^{-1}[\{y_i\}] \subseteq X$, $i=1,2$. $F_1$ and $F_2$ are disjoint closed (by continuity of $f$) subsets of $X$, which is compact and Hausdorff and thus $T_4$ or normal. So in $X$ we can find open disjoint subsets $U_1, U_2$ such that $F_1 \subseteq U_1$ and $F_2 \subseteq U_2$.

Now define $O_i = Y\setminus f[X\setminus U_i]$ for $i=1,2$.

These sets are open as $f$ is a closed map and complements of open sets are closed and complements of closed sets are open.

For each $i$ $y_i \in O_i$: suppose $y_i \notin O_i$ then by definition $y_i \in f[X\setminus U_i]$ so $y_i = f(p)$ for some $p \in X \setminus U_i$. But by definition $p \in F_i$ as it maps to $y_i$ and so should be in $U_i$ by $F_i \subseteq U_i$, a contradiction with $p \in X \setminus U_i$. This contradiction shows that $y_i \in O_i$.

$O_1 \cap O_2 = \emptyset$. Suppose that $y$ lies in both. Then for some $x \in X$, $f(x) = y$. As $U_1$ and $U_2$ are disjoint, $x$ must lie in $X\setminus U_1$ or in $X \setminus U_2$ (or both); suppose it lies in $X \setminus U_j$ for some $j \in \{1,2\}$. But then $y = f(x) \in f[X\setminus U_j]$ which means $y \notin O_j$, which is our contradiction.

So $O_1$ and $O_2$ are the required disjoint open neighbourhoods of $y_1$ resp. $y_2$.

Sanity check: we used the closedness of $f$ and its Hausdorffness too, (also to get normality from the combination with compactness), and the continuity of $f$ and its surjectivity. The proof would also have worked if $X$ would have been merely $T_4$ (normal and $T_1$) and $f$ the same (continuous closed surjection).

$\endgroup$
0
$\begingroup$

Let $y_1 \neq y_2$ in Y. Pick $x_1 ,x_2$ such that $f(x_1)=y_1$ and $f(x_2)=y_2$ Claim: there exist neighborhoods U and V of $x_1$ and $x_2$ such that $f(U) \cap f(V)$ is empty. [ This is where continuity is used]. If not, for any neighborhoods U and V there exist points $t_{UV}$, $s_{UV}$ in U and V such that $f(t_{UV})=f(s_{UV})$ Order the pairs (U,V) by $(U,V) \geq (S,T)$ if U is a subset of S and V is a subset of T. We get nets $\{t_{UV}\}$ and $\{t_{UV}\}$ converging to $x_1$ and $x_2$ respectively. Since f is continuous and $f(t_{UV})=f(s_{UV})$ we get $y_1=f(x_1)=f(x_2)=y_2$ which is a contradiction. This proves the claim. Next pick open sets A, B containing $x_1$ and $x_2$ such that their closures are contained in U and V respectively. Now the complements of the images of the closures of A and B are disjoint neighborhoods of $y_1$ and $y_2$.

$\endgroup$

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.