Talk:Metric space

Quotient metric spaces

"Moreover, if M is a compact space, then the induced topology on M/~ is the quotient topology." — I guess, a condition is forgotten: each equivalence class must be closed. Boris Tsirelson (talk) 18:01, 2 October 2015 (UTC)

Oops, no, the situation is much worse; see quotient of metric spaces on MathOverFlow. Namely, if we collapse (to a point) each closed interval complementary to the Cantor set, then the topological quotient space is homeomorphic to [0,1], but the pseudometric defined in the article is trivial (all distances are 0). Boris Tsirelson (talk) 19:33, 2 October 2015 (UTC)

The statement quoted above was written by Fiedorow on 4 December 2005 at 18:07. He is not active since Oct 2012. Boris Tsirelson (talk) 19:45, 2 October 2015 (UTC)

I delete the problematic statement. Boris Tsirelson (talk) 20:02, 2 October 2015 (UTC)

SNCF vs post office metric

The SNCF and the post office metric are two distinct concepts. The distance between Marseille and Lyon with respect to the SNCF metric is just the Euclidian distance bewtween the two cities, whereas it is the sum of the Euclidian distances Marseille-Paris and Lyon-Paris with respect to the post office metric. 129.13.72.198 (talk) 15:12, 22 April 2016 (UTC)

A statement about closure

Need proof that $\operatorname {cl} (A\cup B)=\operatorname {cl} A\cup \operatorname {cl} B$ in metric spaces. — Preceding unsigned comment added by VictorPorton (talk • contribs) 12:54, 15 November 2019 (UTC)

@VictorPorton:

For $\Longleftarrow$: every point of $A$ belongs to $C=A\cup B$, so every sequence of points from $A$ is a sequence from $C$. Hence any limit point of $A$ is also a limit point of $C$. Similarly for $B$.

For $\Longrightarrow$: every sequence from $C$ contains an infinite subsequence from $A$ or an infinite subsequence from $B$ (or both), so any limit point of $C$ is a limit point of $A$ or of $B$ (or both).

CiaPan (talk) 13:45, 15 November 2019 (UTC)

Diameter of a bounded set

The image provided seems to suggest that the diameter of a bounded set is the diameter of the smallest ball which contains it, which is incorrect (e.g. in $\mathbb {R} ^{2}$ a unit equilateral triangle has diameter $1$ but the smallest containing ball has diameter ${\frac {2}{\sqrt {3}}}$). What should be done with it? Boboquack (talk) 07:44, 22 January 2020 (UTC)

@Boboquack: The best way IMHO will be preparing a correct image without a circle and uploading it as a newer version at Commons:File:Diameter of a Set.svg (the link 'Upload a new version of this file' in the 'File history' section, below the table of versions). --CiaPan (talk) 08:42, 22 January 2020 (UTC)

@CiaPan: I don't have the capacity to do that easily on my device, unfortunately. Boboquack (talk) 02:22, 23 January 2020 (UTC)

@Boboquack: I have uploaded an updated file. CiaPan (talk) 10:02, 23 January 2020 (UTC)  Done

Problem in section on Generalizations of metric spaces

This section talks about relaxing axioms 1 to 4, but there are only 3, as they've been refactored. I don't think I know how to correct it, given that removing eg 2 or 3 may mean you have to add positivity back in. akay (talk) 09:20, 1 February 2021 (UTC)

I have tried to reword the section Metric space § Generalizations of metric spaces to remove explicit reference to the number of axioms. Please check that I didn't mess it up.

FWIW, I would prefer to go back to four axioms, even if they are redundant. It's a very small point, but I believe explicitly stating non-negativity early on really helps the reader on their first encounter with the definition. – Tea2min (talk) 10:52, 1 February 2021 (UTC)

Notification

the propostions : 2 d(x,y)>=0 and d(x,y)>=0 are equivalent — Preceding unsigned comment added by Boutarfa Nafia (talk • contribs) 22:15, 9 February 2022 (UTC)

Link to page "Metric (mathematics)"

I have noticed that this page does not provide any link to the page Metric (mathematics), instead rendering all instances of the term in boldface, as if "Metric (mathematics)" redirects to this page. Should we consider merging the two, or should we link between them? 110521sgl (talk) 13:30, 26 August 2022 (UTC)

I'm currently working on merging the two as well as rewriting this article. There's a discussion at Wikipedia talk:WikiProject Mathematics#Metric space and Metric (mathematics). Please feel free to help out! --platypeanArchcow (talk) 16:11, 26 August 2022 (UTC)

A bit more info on the rewrite -- I'm using group (mathematics), Hilbert space, and field (mathematics) as models of what to work towards. "Group" is a featured article and the other two are officially Good Articles. --platypeanArchcow (talk) 16:19, 26 August 2022 (UTC)

Wrong definition for metric

Hi! The definition of the metric is false (btw it is also false on the page for distance). What you are defining is a pseudometric when we only request d(x,x) = 0 (which is defined correctly later on in the section on generalizations). A distance has to satisfy the stronger condition that d(x,y) = 0 iff x = y. 69.159.147.230 (talk) 19:23, 24 January 2023 (UTC)

Or rather I should say that it is a very weird /unconventional way of doing it since 1+2 => (d(x,y) = 0 iff x=y) 69.159.147.230 (talk) 19:25, 24 January 2023 (UTC)

I have stumbled upon this difference too (https://us.metamath.org/mpeuni/df-met.html and https://archiveofourown.org/works/777002/chapters/11228542) and have explained the equivalence in the article  AltoStev (talk) 10:50, 15 August 2023 (UTC)

At Most

> For any point x in a metric space M and any real number r > 0, the open ball of radius r around x is defined to be the set of points that are at most distance r from x:

That definition is for a closed ball - it includes points on the boundary, at distance r. For an open ball, you need all points less than distance r. 203.13.3.93 (talk) 01:34, 5 April 2023 (UTC)

Coproducts of pseudoquasimetric spaces

In the section on pseudoquasimetric spaces, the wording currently seems to imply that the category of non-extended pseudoquasimetric spaces still admits all finite coproducts, just not infinite ones. Isn't that false? The distance between elements of X and Y in X+Y would have to be infinite in order for X+Y to admit nonexpansive maps to all spaces X and Y map to, but that's not possible with a non-extended metric, so for any two nonempty spaces there should not be a coproduct at all. Peabrainiac (talk) 21:54, 13 September 2023 (UTC)