Talk:Base (topology)

Better
Dcoetzee:

Overall, I think your changes make the article much better. I just have a couple comments.


 * 1) My main concern is that it's still not clear that you can DEFINE a topology by giving a base. For people who are familiar with the concept of "generating objects" (i.e. basis of a vector space, set generating a free group, etc.) the distinction between the generating object being used to understand or analyse a previously given structure vs using that object to define a new structure seems obvious...but to people not familiar with it, I think it's very confusing at first.  In the article, you talk in several places about bases "generating" topologies, and while we know that you mean "base defining topology", the reader might not.  So, when they read "An example of a collection of open sets which is not a basis is the set S of all semi-infinite intervals of..." my concern is that the reader might ask "not a basis OF WHAT TOPOLOGY"? It was to avoid this confusion in the reader that I explicitly spelled out the two uses of the term. While my presentation was admittedly too formal, I think the distinction still needs to be made explicit.
 * 2) A basis of a vector space is far from unique, either. I can't think of many generating objects in math that are completely unique, so I'm not sure what to compare it to, here.

Otherwise, your additions are very nice. Revolver


 * Oops. A basis of a vector space isn't unique, but has unique cardinality, at least. This is what I meant to say here. Fixing this.


 * As for the "not a basis of what topology", I think it suffices to say "not a basis of any topology on this set". I see how the concept and the two uses could be confusing though, and need to be distinguished, but not as severely as they were in the original article (to say these are the two uses of bases is a bit extreme).
 * Derrick Coetzee 01:32, 2 Jan 2004 (UTC)

A compact space with any topology has a finite base (because any base forms an open cover.) Similarly, a Lindelöf space with any topology has a countable base, and so is second-countable.

Is this true? It doesn't seem right. It's true, any base of a compact set is an open cover, but that doesn't mean it's finite. I'm not certain about the Lindelof thing, I think it's true if you throw in another assumption (Hausdorff, probably, but I don't remember.) Revolver


 * Oops, this of course isn't true. Every base of a compact space has a finite subset that covers, being an open cover, but it need not be a base itself. Same bad reasoning into the second fact. I shall remove these. I think we need more theorems about bases in general, and maybe an example of a proof. I want to get the idea across of why bases are useful, which in my opinion is because it's generally easier to deal with an arbitrary base element than, say, an arbitrary open set (if the basis is nicely chosen of course).
 * Derrick Coetzee 01:32, 2 Jan 2004 (UTC)

Question from a padawan
What's the difference between a base of T and the power set of T?

What's an example of a base that is not the set of all subsets of T?

...In the least number of dimensions possible, please; I am actually very stupid, no matter what my friends say.

Thank you. [curtsies] Verdana ♥ Bold 13:48, 8 March 2016 (UTC)


 * IMHO, talking about "the power set of T" and "the set of all subsets of T" (which are the same) does not make sense in this context. You need to reformulate your questions. --Beroal (talk) 07:15, 29 April 2017 (UTC)

Introduction and definition do not match
The introduction mentions T (B is a base wrt T if the unions of B equal T); while the formal definition doesn't make any reference. TheZuza777 (talk) 19:45, 24 May 2017 (UTC)
 * The formal definition doesn't make any reference to T because it is independent of T. B is a base on its own. "B is a base" is a term. You can read a statement in the article below the definition that B is a base iff the closure of B under unions (which may be named T) is a topology. This is how B is connected to topologies, but this is not a definition of base. "B is a base wrt T" is a different term. --Beroal (talk) 05:55, 25 May 2017 (UTC)

Confusing first phrase
"In mathematics, a base (or basis) B of a topology on a set X is a collection of subsets of X that is stable by finite intersection." Buiquangtu (talk) 05:30, 14 July 2020 (UTC)
 * The collection should cover X
 * The collection needs not be stable by finite intersection. A finite intersection is a union of some of its members, not necessarily one of its members (which is what "stable" means, right?)
 * Thanks for pointing this incoherency between the lead and the body of the article. In fact, there are two different definitions, that of the lead and that of the article. The fact that the usual basis of the Zariski topology satisfies both definitions is probably the reason of this confusion. I have fixed the lead. D.Lazard (talk) 08:59, 14 July 2020 (UTC)

Rewrote intro
Problem with old intro is it was more technical than it had to be. The first line of the old intro was:


 * In mathematics, a base (or basis) $B$ of a topology on a set $X$ is a collection of subsets of $X$ such that every finite intersection of elements of $B$ (including $X$ itself, which is, by a standard convention, the empty intersection) is a union of elements of $B$.

There is no need to mention the nullary intersection convention. The description involving finite intersections of sets in $B$ is more technical than necessary for the intro. This technical description can be given later in the article. Mgkrupa 00:42, 23 October 2020 (UTC)
 * Also, I'd like to see some well-known sources describe the nullary intersection convention as "a standard convention." Mgkrupa  00:45, 23 October 2020 (UTC)

Misleading definition of base
The definition of base for a topology given at the top of the lead (family of open sets such every open set is a union of a subfamily of that family) is the correct one that should be emphasized. This is the standard definition.

Now we (mathematicians) know that such families serving as a base for some topology satisfy two specific properties given at the top of the Definition section and any family with these two properties will be a base for the topology it generates. But the two properties should not be the definition of a base. Note that the unfortunate change that made this the definition of base in the Definition section was introduced by misguided editor on Jan 9, 2016 here (https://en.wikipedia.org/w/index.php?title=Base_(topology)&diff=698955079&oldid=644935215) and nobody has bothered to clean this up since.

FYI, I am planning a slight rewrite to make things clearer. And reduce the potential confusion between "base for a given topology" versus "base for a topology", which could be confusing for a beginner. PatrickR2 (talk) 07:48, 13 April 2022 (UTC)


 * I have updated the section "Definition and basic properties" to emphasize the correct definition as explained above. As I was revising this, I was noticing various other problems/inconsistencies in that section, so I ended up rewriting most of it.  Specifically, the notion of topology generated by a family of sets makes sense even if the family is not a base, so had to rewrite things accordingly.  To keep things tighter I also removed some minutiae, inconsequential statements which, while not logically incorrect, were not adding much usefulness to the article.  I'll let this sit as is for a few days, in case people have comments.  Then clean up a few other things in the article. PatrickR2 (talk) 03:47, 16 April 2022 (UTC)

Additionally, the third paragraph of the lead (as of 27 Apr 2022) makes a big deal of the notion of "base for a topology on X", and cites three references to support this: Bourbaki, Willard, and Dugundji. This is slightly dishonest, as Bourbaki and Willard have absolutely no mention of that terminology. Only Dugundji mentions a collection of sets being a basis for some topology as part of its explanations on pp. 66-67, without raising this to the level of a formal definition. Accordingly, I'll be simplifying the lead slightly. PatrickR2 (talk) 04:38, 28 April 2022 (UTC)


 * In my copy of Willard, page 38 Theorem 5.3 states
 * B is a base for a topology on X iff ...
 * This is Willard, original edition, Addison-Wesley, not the Dover edition, but I can't imagine Dover is any different.


 * I also don't quite understand the stylistic complaint. If a text says "Define A as foo. Theorem: A iff B" then I don't understand why B cannot validly be used as the definition, which is what the Tomtheebomb edit did. The stylistic complaint might only be that if some student cracks open a copy of Willard, and sees definition A (top of page 38) they might complain that it doesn't match wikipedias's B (which doesn't appear until a mere 1/2 page later). 67.198.37.16 (talk) 18:02, 2 December 2023 (UTC)

Please inspect the sentence
Hi, what is the application of "<" in "some<" in this sentence: "For example, each of the following families of subset of $$$\mathbb{R}$$$ is closed under finite intersections and so each forms a basis for some< topology on $\mathbb{R}$"

Thanks, Hooman Mallahzadeh (talk) 13:20, 15 December 2022 (UTC)


 * This is certainly a typo introduced by inadvertently clicking on the corresponding button in the edit menu. To be removed. ✅. D.Lazard (talk) 13:55, 15 December 2022 (UTC)