3
$\begingroup$

A subset $A$ of a uniform space is said to be bounded if for each entourage $V$, $A$ is a subset of $V^n[F]$ for some natural number $n$ and some finite set $F$. A subset of a metric space is said to be bounded if it is contained in some open ball. Now if $U$ is the uniformity induced by a metric $d$, then the open balls with respect to $d$ are entourages in $U$, so clearly a set bounded with respect to $d$ is also bounded with respect to $U$.

But this journal paper says that the converse is not true:

In a metric space $(X,d)$ we have that each set that is bounded for the metric $d$ is bounded ... for the underlying uniformity, but the converse is in general not true.

So my question is, what is an example of a metric space $(X,d)$ where some sets bounded with respect to the the uniformity induced by $d$ are not bounded with respect to $d$?

$\endgroup$
  • $\begingroup$ The paper quotes "Topologie Génerale" as a source for the definition. I looked (it is defined in an exercise, not in the main text) but there is no mention of a comparison with the metric case there. Sloppy of that paper to state this without reference. Have you asked the author? $\endgroup$ – Henno Brandsma Dec 22 '18 at 22:13
  • $\begingroup$ The final part of said exercise suggests that the converse does hold for connected spaces, so a counterexample has to be disconnected I think. $\endgroup$ – Henno Brandsma Dec 22 '18 at 22:34
  • $\begingroup$ Maybe the author means that $X$ can have an equivalent uniformity (ie yields same topology) and be bounded in that uniformity but not in $d$? I think one can prove that if we use the uniformity induced by $d$ then Bourbaki-bounded sets are $d$-bounded. $\endgroup$ – Henno Brandsma Dec 22 '18 at 23:13
  • $\begingroup$ @HennoBrandsma I just emailed the author, let's see what he says. By the way, do you know whether two metrics which induce the same uniformity have the same bounded sets? $\endgroup$ – Keshav Srinivasan Dec 22 '18 at 23:44
  • $\begingroup$ @HennoBrandsma I'm starting to question my whole understanding of uniformities. This book claims that for any metric $d$, the metric $\frac{d}{1+d}$ is uniformly equivalent to $d$. But if two metrics are uniformly equivalent then don't they induce the same uniformity? I'm really confused. $\endgroup$ – Keshav Srinivasan Dec 23 '18 at 1:45
1
+150
$\begingroup$

I see the situation vice versa. Assume that a subset $A$ of a metric space $(X, d)$ is bounded with respect to the uniformity $\mathcal U(d)$ induced by $d$. Pick an arbitrary $\varepsilon>0$. Let $$V=\{(x,y)\in X\times X: d(x,y)<\varepsilon\}\in\mathcal U(d).$$ Therefore there exists a number $n$ and a finite subset $F$ of $X$ such that $A\subset V^n[F]$. That is for each point $y\in A$ there exists a point $x\in F$ such that $y\in V^n[F]$. The triangle inequality implies that $d(x,y)<n\varepsilon$. Pick any point $x\in F$. Then the triangle inequality implies that the set $A$ is contained in an open ball centered at $x$ with the radius $n\varepsilon+\max \{d(x,y):y\in F\}$, that is $A$ is bounded with respect to the metric $d$.

Conversely, let $X$ be an infinite set endowed with the metric $d(x,y)=0$, if $x=y$, and $d(x,y)=1$, otherwise for each $x,y$ in $X$. Then $X$ is contained in an open ball of radius $2$ centered at any point $x\in X$. Let $$V=\{(x,y)\in X\times X: d(x,y)<1\}\in\mathcal U(d).$$ Then $V$ is the diagonal of the set $X\times X$, so $V^n=V$ for each $n$. Therefore $V^n[F]=F$ for each (finite) subset $F$ of $X$, that is the space $X$ is not $\mathcal U(d)$-bounded.

$\endgroup$
1
$\begingroup$

In a metric space $(X,d)$ bounded can mean three things: there are three bornologies we can talk about:

  1. $\mathcal{U}(d)$-bounded sets (in the Bourbaki sense).
  2. $\mathcal{U}(d)$-totally bounded sets (which Bourbaki calls precompact).
  3. $d$-bounded sets in the standard sense (being contained in a ball).

If $d$ and $d'$ are uniformly equivalent, of course 1 and 2 are the same but 3 can be different wrt $d$ or $d'$. The $\frac{d}{1+d}$ case is an example.

If $d$ and $d'$ are strongly equivalent, for all $i \in \{1,2,3\}$ $(X,d)$ and $(X,d')$ agree on boundedness notion $i$.

Concretely, take $\mathbb{R}$ in the metric $d(x,y) = \min(|x-y|,1)$. Then $A=\mathbb{R}$ is $d$-bounded but not "bounded" (from the uniformity, as in the paper).

$\endgroup$
  • $\begingroup$ Wait a minute, doesn’t the paper say all $d$-bounded sets are $U(d)$-bounded, but the converse need not be true? So how are you giving an example of a set which is $d$-bounded but not $U(d)$-bounded? $\endgroup$ – Keshav Srinivasan Dec 23 '18 at 14:10
  • $\begingroup$ What am I missing? $\endgroup$ – Keshav Srinivasan Dec 23 '18 at 21:25

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.