6

Wikipedia says "every uniform space is completely regular".

How to prove that every uniform space is completely regular?

porton
  • 5,053

1 Answers1

9

It’s fairly trivial if you use the pseudometric definition of uniform space. Let $\mathscr{D}$ be the set of pseudometrics generating the uniform structure. If $U$ is an open nbhd of some $x\in X$, then there are a finite $\{d_1,\dots,d_n\}\subseteq\mathscr{D}$ and an $\epsilon>0$ such that

$$\bigcap_{k=1}^nB_{d_k}(x,\epsilon)\subseteq U\;.$$

Define $$f:X\to\Bbb R:y\mapsto\max_{1\le k\le n}d_k(x,y)\;;$$

$f$ is continuous, and $f(x)=0$. If $y\in X\setminus U$, then $d_k(x,y)\ge\epsilon$ for some $k\in\{1,\dots,n\}$, so $f(y)\ge\epsilon$. Now let

$$g:X\to[0,1]:y\mapsto\frac1\epsilon\min\{f(y),\epsilon\}\;;$$

$g$ is continuous, $g(x)=0$, and $g(y)=1$ for $y\in X\setminus U$.

In any case this is a standard result whose proof can be found in any good topology reference (e.g., Engelking’s General Topology and Willard’s General Topology).

Brian M. Scott
  • 616,228
  • 2
    And how to show that pseudometric definition is equivalent to the classic set of entourages? – porton Sep 16 '13 at 11:25
  • It seems that Engelking’s "General Topology" does not have proofs, only theorems. – porton Sep 16 '13 at 13:21
  • 1
    @porton: Engelking’s General Topology most certainly has proofs. In particular, Corollary $8.1.13$, which is the desired result, is proved. It depends on Cor. $8.1.11$ and Theorem $8.1.10$, both of which are also proved. – Brian M. Scott Sep 16 '13 at 16:35
  • However one must specify that this holds for separating uniform structures. – Francesco Bilotta May 17 '19 at 06:47