In non-Hausdorff topology it is standard to define the Borel algebra of a topological space $X$ as the $\sigma$-algebra generated by the open subsets and the compact saturated subsets. Recall that a subset is saturated if it is an intersection of open subsets, and that compact saturated subsets play the role of compact subsets when the space $X$ is not $T_1$ (which is typically the case for a partially ordered set equipped with the Scott topology for instance).

In this situation, for a continuous function $f : X \to Y$ between topological spaces, is $f$ necessarily measurable?

This question is equivalent to the following. If we write $\uparrow y$ for the intersection of all open subsets containing $y$, which happens to be compact saturated, is it true that $f^{-1}(\uparrow y)$ is measurable for all $y \in Y$?

Thank you very much for your help. Paul

Edit: this question has now been migrated to MathOverflow, see here.

  • $\begingroup$ Inverse image under a continuous map preserves openness (definitionally) and inverse image under a function preserves arbitrary intersections, so the only remaining criterion is properness (inverse image of a compact set being compact). Given that continuous functions aren't usually proper, I expect this not to be the case. $\endgroup$ – jdc May 24 '14 at 20:57
  • $\begingroup$ Dear jdc, since indeed the inverse image of a compact by a compact needs not to be compact, I do not expect this to happen. Without being compact, the inverse image of a compact saturated subset could still be measurable. $\endgroup$ – polmath May 25 '14 at 10:18
  • $\begingroup$ Not to be nitpicky, but migration is when the question is moved. What you did is to crosspost the question on MathOverflow. $\endgroup$ – Asaf Karagila Sep 26 '14 at 12:58
  • $\begingroup$ @AsafKaragila: thanks. Indeed I asked for "official" migration" but the post was too old for that; talking about "crossposting" is certainly more correct. $\endgroup$ – polmath Sep 26 '14 at 13:19

Let $X$ be a topological space with a compact, saturated, non-open set $S$ and $Z$ a space that is not itself the union of countably many compact subsets. Consider the projection $X \times Z \to X$. I suspect the inverse image $S \times Z$ will not behave as hoped. I think it would be as you wanted under the product Borel algebra but not under that of the Borel algebra associated to the product topology.

Or I'm missing something.


I would like to complete my question with the following elements.

If $Y$ is $T_1$, then every subset $\uparrow y$ is closed (it coincides with the singleton $\{ y \}$ which itself coincides with its closure), so the continuous map $f : X \to Y$ is measurable.

If $Y$ is first-countable, then every subset $\uparrow y$ can be written as a countable intersection of open subsets, so again $f$ is measurable.

If $f$ is open and bijective, one can show that the inverse image of $\uparrow y$ is of the form $\uparrow x$, $x \in X$, so $f$ is measurable.

Do we know other such situations? (sufficient conditions on $X$, $Y$ or $f$ in order for $f$ to be measurable?)


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.