MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

An arbitrary union, or a finite intersection, of open sets in a topological space is again open. What name is given to the hypothetical property that an arbitrary intersection of open sets is open?

As an example, consider a partially ordered set $X$. Call a subset $U\subseteq X$ open if $y\le x\in U$ implies $y\in U$. (Bonus question: Are there other interesting examples?)

share|cite|improve this question
It suffices in your example for the order relation to be only a pre-order. That is, you can allow x <= y <= x for distinct x, y. With this addition, the property is fully equivalent. Define x <= y if x is in every open set that y is in. – Joel David Hamkins Mar 16 '10 at 16:14
Right; I noticed that in the Wikipedia article linked to from the answer. – Harald Hanche-Olsen Mar 16 '10 at 16:32
Does "arbitrary" intersection should include "empty" intersection? Then the whole space must be "open". Similarly, arbitrary union means the empty set is "open". This is what I would call a "complete lattice of sets". – Gerald Edgar Mar 16 '10 at 16:33
@Gerald: Yes. But the whole space, and the empty set, are already open per the definition of topology. Surely, “complete lattice of sets” is correct, but not so good when I wish to emphasize the topology aspect. So I'll stick with Alexandrov space. – Harald Hanche-Olsen Mar 16 '10 at 16:44
Harald: the reason I asked was that your original example usually does not have the whole space open. – Gerald Edgar Mar 16 '10 at 17:43
up vote 14 down vote accepted

Alexandrov spaces.

share|cite|improve this answer
Well, that was fast. Thanks. – Harald Hanche-Olsen Mar 16 '10 at 16:12
Alexandrov topology and Alexandrov-discrete space are OK, but "Alexandrov spaces" is mostly used for something else... – Anton Petrunin Mar 16 '10 at 17:40
Peter May uses ‘Alexandrov space’ (really, ‘A-space’): – L Spice Mar 31 '10 at 1:22

Harald, as Peter May observes in his notes on finite topological spaces (but as must already be standard), your pre-ordered sets provide the only examples. Indeed, if $X$ is an Alexandrov space (or Alexandrov-discrete space, whatever the terminology is), then we may impose a pre-order on it by demanding for $x, y \in X$ that $x \le y$ if and only if $x$ lies in every neighbourhood of $y$. This becomes a genuine partial order exactly when $X$ is $T_0$.

(Sorry; I didn't realise until after posting that Joel had already said this.)

share|cite|improve this answer
In general, this is not a partial order, but only a partial pre-order, since it may be that x and y are distinct, but lie in all the same open sets together. But with pre-orders, what you say is true (and was observed previously in the comments to the question). – Joel David Hamkins Mar 31 '10 at 1:33
Joel, thanks; I'm sorry that I missed your comment. I also forgot that we weren't assuming the spaces were $T_0$. Should I delete this answer? – L Spice Mar 31 '10 at 1:38
Oh, its no problem. Probably its fine to keep the answer, if you correct it to explain about the pre-order business. But I think this issue also appears on the Wikipedia page. – Joel David Hamkins Mar 31 '10 at 1:45

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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