Talk:Coherent topology

Slight error
It's a good article, but there is a slight error in there I believe. The discrete topology isn't coherent with every collection of subsets. One implication holds of course. If U (as a subset of X) intersected with A for every A in some collection {A} is open in A then U is open i X (because every U is open i X with the discrete topology). The converse doesn't hold. If U is open in X (which it always is) then the intersection of U with A is not necessarily open in every A in {A}.

YohanN7 (talk) 15:56, 2 May 2008 (UTC)


 * There is no error. Recall that every subset of a discrete space is discrete. So U is open in every A. -- Fropuff (talk) 17:56, 2 May 2008 (UTC)


 * You can't assume that every subset A of X in a collection of subsets has the subspace topology. That is not part of the definition of a coherent topology (the definition that says shortly U is open in X if and only if U intersected with A is open i A for every A in the collection). The subsets are allowed to have a topology of their own! Suppose that the collection of subsets consists of a single set A with the trivial topology. Let A = {a,b}. Then set set {a} is open in X, but not in A.YohanN7 (talk) 17:50, 5 May 2008 (UTC)


 * The article states that a discrete space is coherent with every family of subspaces not subsets, so there is no error. As far as I am aware the coherent topologies are defined with respect to subspaces not with respect to arbitrary inclusions. Perhaps I am mistaken. Can you provided a reference for the more general definition? -- Fropuff (talk) 00:20, 6 May 2008 (UTC)


 * Ok, then what we are discussing is not "right or wrong", but probably merely what one takes for the definition of a coherent topology. My primary reference is actually Willards book referenced in the main article (an excellent book in my oppinion). Choherent topologies are introduced in a problem in chapter 9. I don't have my book at hand at the moment, but i'm fairly sure he uses the word subset instead of subspaces. On the other hand, his convention is to use subset (without further mention of the topology) to mean subspace in the more conventional sense (where the subspace topology is assumed implicitely). So maybe I am the one that is mistaken.


 * The definition that I (incorrectly) assumed from the start was the one where the subsets are free to have any topology (, and in addition are not required to cover X) does make sense too - it becomes a generalization of the one meant in the article. There are examples: Consider a finite space X with the trivial topology and a collection of subsets which are equal to X as sets, but endowed with some sort of Sierpinski topology. For instance let X = {a,b}, and let one subset have the topology {X,0,{a}}, and another the topology {X,0,{b}}. Then the trivial topology is (the unique) coherent with thos subsets.


 * Of course, in most typically occuring examples the two definitions coincide. Best regards! YohanN7 (talk) 17:10, 6 May 2008 (UTC)

About the topological union
Hi,

About the part under the topological union when you say "Let {Xα} be a family of (not necessarily disjoint) topological spaces such that the induced topologies agree on each intersection Xα ∩ Xβ. [...] The inclusion maps will then be topological embeddings and X will be coherent with the subspaces {Xα}."

Is this true? I tried proving it but without success. Then I found in a book an exercise that asked to prove the same thing but with the additional hypothesis that Xα ∩ Xβ must be closed (or open) in Xα and Xβ. And this I proved without trouble.

And while we're at it, I've got another remark/suggestion.

I do not think the term "topological union" is very popular. More so is the term "space generated by a family of topological spaces", and it agrees with other wiki article. For instance the articles "compactly generated space" and "finitely generated space", etc. Quasar987 (talk) 05:13, 12 February 2009 (UTC)


 * The article, as currently written, makes clear that Xα ∩ Xβ must be closed as a part of the definition; so I think that takes care of your first comment. 67.198.37.16 (talk) 16:10, 23 July 2016 (UTC)

Problem understanding article
The article currently says:


 * Given a topological space X and any family of subspaces C there is unique topology on X which is coherent with C. This topology will, in general, be finer than the given topology on X.

I don't understand what this means. My guess is that it means:

Given a set X and any family of subsets C there is a unique topology on X which is coherent with C.

But in this case, what does the "finer" bit mean? It seems to imply that the coherent topology is finer than any given topology on X, i.e. it would always be the discrete topology. This is clearly not true. Can anyone clear up my confusion? 121.98.145.129 (talk) 20:17, 26 November 2009 (UTC)


 * Yes, that wording is awkward. I guess that it is trying to suggest that maybe there are many cases where there is an X that has some natural topology, but that topology is coarser than the coherent topology. However, your guess is not right: the definition (as currently written) requires that X be a topological space (to begin with!) (and not a set)  and that the C's are subspaces (not subsets) that cover X.  No mention of sets anywhere in the definition. Since one is already starting out with X a topological space, the suggestion is that the coherent topology is finer than whatever X "might have started with" (although that is vague, as currently written).  Perhaps one could have an alternative definition with sets, but the current definition seems to be trying to avoid sets, so as to be "pointless". Firming this up would be nice. 67.198.37.16 (talk) 16:13, 23 July 2016 (UTC)

Properties section: second bullet
The second bullet in the list of Properties:
 * If $$D$$ is a refinement of $$C$$ and each $$C_{\alpha}$$ is determined by the family of all $$D_{\beta}$$ contained in $$C_{\alpha}$$ then $$X$$ is determined by $$D.$$

This may need some extra hypothesis, as in general if $$D$$ is a refinement of $$C$$, it is not necessarily the case that the family of all $$D_{\beta}$$ contained in $$C_{\alpha}$$ is a cover of $$C_{\alpha}$$. There may not even be any such $$D_{\beta}$$ contained in $$C_{\alpha}$$.

By the way, I am curious, do you know of any situation where this bullet item is used in the proof of something interesting? PatrickR2 (talk) 00:11, 13 February 2023 (UTC)