1

The Niemytzki plane, also known as the tangent disk space, is a completely regular $T_0$-space that is not normal but is locally metrizable (for its local metrizability, see https://math.stackexchange.com/a/4412135/32337).

What other examples are there of such spaces?

In view of the Urysohn metrization theorem, it suffices to obtain spaces that are completely regular $T_0$ and non-normal and, in addition, locally second-countable.

My purpose in asking the question is to find examples other than the Niemytzki plane that may be used in the construction given in https://math.stackexchange.com/a/4411293/32337, which leads to a non-normal, completely regular $T_0$ space that is a quotient of a normal $T_1$-space.

murray
  • 778
  • 3
  • 16

1 Answers1

1

An interesting example is the subspace $X=A\cup B$ of the Niemitzki plane, where $A=\Bbb R\times \{0\}$ and $B=\{(a/n, 1/n):a\in\Bbb Z\land n\in\Bbb N\}.$ The subspace $A$ is closed in $X$ & discrete, with $|A|=2^{\aleph_0}.$ ... $B$ is a countable union of open singletons of $X$, so $B$ is also discrete. $B$ is dense in $X,$ so $X$ is not normal by the Jones Lemma. Any nbhd in $X$ of any $(r,0)\in A$ has an open-and-closed sub-nbhd $V=\{(r,0)\}\cup W$ where $W\subset B$ and $\overline W=V,$ so $V$ is homeomorphic to the real subspace $\{0\}\cup \{2^{-j}:j\in\Bbb N\}.$

In General Topology by R. Engelking, an example non-normal $Y=A\cup \Bbb N \subset \beta \Bbb N$ is given, where $A$ is a closed discrete subspace of $Y,$ with $A\subset \beta \Bbb N\setminus \Bbb N$ and $|A|=2^{\aleph_0}.$ I noticed that $Y$ is homeomorphic to $X$ of my 1st paragraph.

It seems the hypotheses of the Q are redundant. Wouldn't a locally metrizable space be $T_1$ and $T_{3\frac 1 2}$?

It would be interesting to see an example (if there is one) that does not involve the Jones Lemma.

BTW, in the Niemietzki plane, a basic open set $\{(r,0)\}\cup D,$ where $D$ is an open disc in $\Bbb R^+\times \Bbb R,$ of radius $s>0$ and centered at $(r,s),$ can also be seen to be metrizable by another strong metrization theorem: A $T_1,T_3$ space is metrizable iff it has a $\sigma$-locally finite base (basis). From this theorem, if $D$ is a metrizable open subspace of $E=D\cup \{p\}$ where $E$ is $T_1, T_3$ and $p$ has countable character in $E$, then $E$ is metrizable.

  • "Wouldn't a locally metrizable space be $T_1$?" Not necessarily, e.g. the "line with two origins". – David Hartley May 01 '22 at 19:23
  • @DavidHartley. Right. Forgot about that. – DanielWainfleet May 01 '22 at 19:30
  • @DanielWainfleet: can you point me to the location of that example in Engelking? – murray May 02 '22 at 19:51
  • Engelking 3.6.18 and 3.6.19. On re-reading, I think 3.6.19 can be simplified: $X=\Bbb N\cup {x_t:t\in I}$ where ${x_t:t\in I}$ is a discrete subspace of $\beta \Bbb N$ \ $\Bbb N.$ (...$I=[0,1]...)$ He cites Theorem 3.3.9 to show that $X$ \ $\Bbb N$ is closed in $X.$ But $\Bbb N$ is open in any compactification of $\Bbb N$ because $\Bbb N$ is locally compact, so if $\Bbb N\subset X\subset \beta\Bbb N$ then $\Bbb N$ is open in $X$.... BTW he also cites 2.1.10, which is that a separable normal space cannot have a closed discrete subspace of cardinality $c$. – DanielWainfleet May 03 '22 at 05:46