0
$\begingroup$

I am looking for some proof insight or literature references for a statement which, if it's actually true, is probably a pretty trivial thing. I hope the question has not been asked here before, and that my notation is not too strange. Any help is appreciated!

Let $X, Y$ be topological spaces. Let $A$ be the collection of open sets of $X$, let $B$ be the collection of open sets of $Y$, and let $B^*$ be a base for $Y$. Let $F^*: B^* \to A$ be a function such that

  1. $F^*(Y) = X$.
  2. $F^*(\emptyset) = \emptyset$.
  3. $F^*(U\cap V) = F^*(U) \cap F^*(V)$ for $U, V \in B^*$.

Question: Can we extend $F^*$ to a function $F : B \to A$ that commutes with arbitrary unions and finite intersections? Is this extension unique?

Stated from a different perspective, is there a continuous function $f: X \to Y$ such that $f^{-1}(U) = F^*(U)$ for each $U$ in $B^*$? If it exists, how do we define it on individual points of $X$? Is $f$ unique?

What I see so far: I can think of at least a partial answer to my questions.

First note that by definition, $F^*$ preserves finite intersections from $B^*.\ $ For $U_i$ an arbitrary family of open sets in $B^*$, define $F(\cup U_i) = \cup F^*(U_i)$. This extends $F^*$ to all of $B$, and obviously commutes over arbitrary unions. To see it commutes over finite intersections let $U_i, V_j$ be arbitrary families of open sets in $B^*$ and consider

$$F((\cup U_i)\cap(\cup V_j)) = F(\bigcup_i \bigcup_j U_i \cap V_j) = \bigcup_i \bigcup_j F^*(U_i)\cap F^*(V_j)$$ $$=(\cup F^*(U_i)) \cap (\cup F^*(V_j))$$ $$=F(\cup U_i)\cap F(\cup V_j)$$

How can we decide whether $F$ is unique? And can we take F at large an use it to construct $f$'s action on individual points of $X$?

$\endgroup$
2
  • $\begingroup$ D'oh. If $Y$ has the indiscrete topology there is only one admissible $F^*$. But in that case, any function from $X$'s carrier set to $Y$ is going to be continuous. I guess I'd like to pivot the question and also ask whether there are conditions that let us uniquely define $f$. I'm interested in particular in a case where $X$ is normal and $Y = \mathbb{R},$ or where $X$ or $Y$ (or both) are sober. $\endgroup$
    – Max Suica
    Sep 7, 2014 at 9:00
  • $\begingroup$ It is not always possible, if we manage to construct a map $F:B^* \to A^*$ with the properties $(1)-(3)$ above but which doesn't respect arbitrary intersections (only finite ones). Then there can't be a map $f$ with the desired properties, because $f^{-1}:\mathcal{P}(Y) \to \mathcal{P}(X)$ respects arbitrary intersections. $\endgroup$ Sep 9, 2014 at 11:29

1 Answer 1

1
$\begingroup$

The answer to the general question

is there a continuous function $f:X\to Y$ such that $f^{-1}(U)=F^*(U)$ for each $U\in B^*$

is "No". Consider the following example:

Let $X=Y=\mathbb{N}$ and let $L = \{S \subseteq \mathbb{N}: 1\in S \textrm{ and }\mathbb{N}\setminus S \textrm{ is finite}\}$.

Put $A = B^* = L \cup \{\emptyset\} \cup \{\{1\}\}$. It is easy to verify that $A= B^*$ is a topology (so in particular, $B^*$ is a base for the topology $B^*$). Let $F^*:B^*\to A$ be defined by

  • $F^*(S) = S$ for all $S\in L$;

  • $\{1\} \mapsto \emptyset$ and $\emptyset \mapsto \emptyset$.

Then $F^*$ obeys the conditions 1. to 3. in the question, but note that $F^*$ does not commute with arbitrary intersections, because $F^*(\bigcap L) = F^*(\{1\}) = \emptyset \neq \{1\} = \bigcap F^*(L)$. Now, for any $f:X\to Y$ the map $f^{-1}(\cdot)$ commutes with arbitrary intersections. So there cannot be $f:\mathbb{N}\to\mathbb{N}$ such that $f^{-1}(U) = F^*(U)$ for all $U\in B^*$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.