Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I've looked around but all I could find is that if X is separable then X has ccc. Can anybody give me some help?

share|improve this question
add comment

3 Answers

up vote 5 down vote accepted

Let $X$ be a (non-empty) c.c.c. metric space. Let us temporarily fix a positive integer $n\geq 1$.

Let $\mathscr{F}_n$ be the family of classes of open balls of radius $1/n$ (in $X$) which are pairwise disjoint. In other words, an element of $\mathscr{F}_n$ is a class $\mathcal{C}$, whose elements are open balls, say $B^\mathcal{C}_i$, of radius $1/n$, and such that $B^\mathcal{C}_i\cap B^\mathcal{C}_j=\varnothing$ if $i\neq j$.

We can partially order $\mathscr{F}_n$ by inclusion. It's an easy exercise to verify that every chain in $\mathscr{F}_n$ has an upper bound (given, obviously, by the union of the elements of the chain) - and an even easier one to show that $\mathscr{F}_n\neq\varnothing$- . By Zorn's Lemma, $\mathscr{F}_n$ has a maximal element, say $\mathcal{C}_n$. Since open balls are open and $X$ is c.c.c., $\mathcal{C}_n$ is enumerable. Also, since it's elements are balls, we can find point $x^n_1,x^n_2,\ldots\in X$ such that $$\mathcal{C}_n=\left\{B_{1/n}(x^n_1),B_{1/n}(x^n_2),\ldots\right\}$$ (where $B_r(x)=\left\{y\in X:d(x,y)<r\right\}$, as usual)

Now, we can do the above procedure for each positive integer $n\geq 1$, thus finding maximal classes $\mathcal{C}_n$ of pairwise disjoint balls of radius $1/n$ and centers $x^n_1, x^n_2,\ldots$ Now, let $D$ be the set of all such $x^n_i$. Then $D$ is obviously enumerable, and we claim that it is dense in $X$.

Indeed, if it weren't, we could find a point $y\in X$ and a positive integer $N\geq 1$ such that $B_{2/N}(y)\cap D=\varnothing$. Hence, for all $x\in D$, we have $B_{1/N}(x)\cap B_{1/N}(y)=\varnothing$. Now, we can consider the set $C'_N=\left\{B_{1/N}(y)\right\}\cup C_N$. It is obvious then that $C'_N\in\mathscr{F}_N$ and $C'_N$ is strictly larger than $C_N$, contradicting it's maximality.

We can then conclude that $D$ is an enumerable dense set in $X$ which is therefore separable.

share|improve this answer
add comment

A proof not using $\sigma$-discrete bases: let $(X,d)$ be a metric space with ccc. Then for each $n \in \mathbb{N}$, let $D_n = \{ B(x, \frac{1}{n}) : x \in C_n \}$, be a maximal (by inclusion: we cannot enlarge this family) disjoint family of open balls in $X$, all of radius $\frac{1}{n}$. The set $C_n$ is here (by definition) the set of centres of these balls. A simple Zorn's lemma application shows that such a family exists, for each $n$. As $X$ has ccc, each $C_n$ is a countable set.

I claim that $D = \cup_n D_n$ is dense in $X$. To this end, we show that $D$ intersects each ball $B(x,\frac{1}{k})$. Suppose not. Then in particular, for each $y \in C_{2k} \subset D$, $d(x,y) \ge \frac{1}{k}$, which implies that the ball $B(x, \frac{1}{2k})$ is disjoint from all balls $B(y, \frac{1}{2k})$, where $y \in C_{2k}$, and this in turn means that we could have extended $D_{2k}$ by adding the ball $B(x, \frac{1}{2k})$ to it, contradicting the way we chose the $D_n$. This contradiction completes the proof, as $D$ is a countable dense set in $X$.

share|improve this answer
add comment

Let $X$ be a ccc metric space. Since $X$ is metric, it has a $\sigma$-discrete base $\mathscr{B}=\bigcup_{n\in\omega}\mathscr{B}_n$. In particular, each $\mathscr{B}_n$ is a pairwise disjoint family of open sets, hence countable, and $\mathscr{B}$ is therefore also countable. Thus, $X$ is second countable and hence separable.

More generally, if $c(X)=\kappa$, then each of the countably many discrete families of a $\sigma$-discrete base for $X$ has cardinality at most $\kappa$, so $w(X)\le\kappa$. On the other hand, it’s always true that $c(X)\le w(X)$, so if $X$ is metric we always have $c(X)=w(X)$. And since

$$c(X)\le d(X)\le hd(X)\le nw(X)\le w(X)$$

in any space, we get equality of cellularity, density, hereditary density, net weight, and weight in metric spaces.

share|improve this answer
add comment

Your Answer

 
discard

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.