Let $$\mathcal{B}_0=\{[a,b):a,b\in\mathbb{R},a<b\}$$ and $$\mathcal{B}_1=\{[a,b):a\in\mathbb{Q},b\in\mathbb{R},a<b\}.$$

Let $\sigma$ and $\tau$ be the topologies over $\mathbb{R}$ generated by the basis $\mathcal{B_0}$ and $\mathcal{B_1}$, respectively.

Are these spaces metrizable?

a) $(\mathbb{R},\sigma)\times(\mathbb{R},\sigma)$

b) $(\mathbb{R},\sigma)\times(\mathbb{R},\tau)$

c) $(\mathbb{R},\tau)\times(\mathbb{R},\tau)$

Well, for the part $a)$, $(\mathbb{R},\sigma)$ is just the Sorgenfrey line and it is easily shown that $(\mathbb{R},\sigma)\times(\mathbb{R},\sigma)$ is not normal by Jones' lemma. Hence this space cannot be metrizable because every metric space is normal.

How about b) and c)? Could you give me any hint?

I can use some metrizations theorems. For example Urysonh theorem (regular + second countable implies metrizable) and Nagata-Smirnov (though I think Nagata-Smirnov is a too heavy theorem).

Thank you all.


1 Answer 1


For (b): $R_1=(R,\sigma)$, which is not metrizable, is homeomorphic to the subspace $R_1\times \{0\}$ of the space $R_1\times (R,\tau)$. A metrizable space cannot have a non-metrizable subspace......For (c), if $x\in Q$ let $B_x= \{[x,x+q):q\in Q^+\}.$ If $x\in R\backslash Q$ let $B_x=\{[q_1,q_2):q_1,q_2\in Q\land q_1<x<q_2\}.$ Then in $(R,\tau)$ the family $B_x$ is a local base at $x$ and every member of $B_x$ is closed. So $(R,\tau)$ is totally disconnected and is therefore a $T_{3\frac {1}{2}}$ space.From the Urysohn theorem it is metrizable. Hence its square is metrizable....... Remark : Another way to show that the Sorgenfry line is not metrizable: If ($X,d)$ is a metric space and $D$ is a dense subset of $X$ then $\{B_d(p,q) :p\in D\land q\in Q^+\}$ is a base. So a separable metrizable space is second-countable. Now $Q$ is dense in $ R_1=(R,\sigma)$ but if $B$ is any base for $R_1,$ then for $x\in R$ let $y_x>x$ and $f(x)\in B$ such that $[x,y_x)\subset f(x)\in B.$ Then $f:R\to B$ is 1-to-1 (because $\min f(x)=x$) so $B$ is uncountable.

  • $\begingroup$ This post needs some "air" ... $\endgroup$
    – Déjà vu
    Jan 3, 2016 at 6:00
  • $\begingroup$ @ringe . What do you mean ? $\endgroup$ Jan 3, 2016 at 6:01
  • $\begingroup$ @user254665 I see, thank you. It seems that the same argument can be given for part a), so we don't have the need to use Jones' Lemma. Also, now that I think, for the part c) we would only need to think about if $(\mathbb{R},\tau)$ is metrizable or not. It's second countable (a basis is family of sets $[q,q+1/n),q\in\mathbb{Q}$). But could it be regular? $\endgroup$
    – JonSK
    Jan 3, 2016 at 6:16
  • $\begingroup$ $(R,\tau)$ is $T(3\frac {1}{2})$ so it is $T(3)$. And it's obviously Hausdorff. And $\{[q,q') :q,q'\in Q\land q<q'\}$ is a countable base $\endgroup$ Jan 3, 2016 at 6:39
  • $\begingroup$ I just realized that JonSK 's last comment, above, must have arrived before I added the answer to (c). $\endgroup$ Jan 3, 2016 at 7:08

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.