3
$\begingroup$

Given a poset $(P,\leq)$ the interval topology on $P$ is generated by $$\{P\setminus\downarrow x : x\in P\} \cup \{P\setminus\uparrow x : x\in P\},$$ where $\downarrow x = \{y\in P: y\leq x\}$ and $\uparrow x = \{y\in P: y\geq x\}$.

Let $\{P_i : i\in I\}$ be a family of posets such that the interval topology of each $P_i$ is Hausdorff. Is the interval topology of $\prod_{i\in I} P_i$ also Hausdorff? (We say that $(x_i)_{i\in I} \leq (y_i)_{i\in I}$ in $\prod_{i\in I} P_i$ if and only if $x_i \leq y_i$ for all $i\in I$.)

$\endgroup$

2 Answers 2

5
$\begingroup$

The statement in the answer by Dominic that the interval topology of the product poset is the product topology of the interval topologies is incorrect. The argument that the product topology contains the interval topology is correct, but the one for the opposite containment is not, as shown in this MO answer.

And in fact the interval topology for $\mathbb{R} \times \mathbb{R}$ is not Hausdorff, because as shown in the linked answer, every basic open $U$ in the interval topology contains a "northwest quadrant", i.e., a subset of the form $NW_{a b} = \{(x, y): x < a\; \text{and}\; y > b\}$, and the intersection of two northwest quadrants is always nonempty because $NW_{a b} \cap NW_{a' b'} = NW_{\min(a, a'), \max(b, b')}$.

$\endgroup$
3
  • $\begingroup$ But why is every subbasic set of the product topology $\tau_p$ open in the interval topology (see my (non-)answer, starting from "Conversely") $\endgroup$ Nov 22, 2017 at 12:29
  • $\begingroup$ My linked answer says that subbasis elements like $\pi^{-1}(U)$ are not open in the interval topology, because $\pi_1^{-1}(U)$ is too small to contain any union of a northwest quadrant and a southeast quadrant (and any basic in the interval topology contains such a union). In brief, your displayed formula for $U$ is incorrect. Try drawing a picture of a $P \setminus (\uparrow (x, y))$, which looks like a complement of a northeast quadrant (i.e., looks like a "thick L-shape"), and you should see this easily. $\endgroup$
    – Todd Trimble
    Nov 22, 2017 at 12:43
  • $\begingroup$ That's right, thanks for the picture! $\endgroup$ Nov 22, 2017 at 12:44
2
$\begingroup$

The answer is Yes, because the product topology of Hausdorff spaces is Hausdorff again, and because the product topology of the interval topologies of a family of posets equals the interval topology of the product:

Lemma. The interval topology $\tau_i = \tau_{\textrm{int}}(\prod_{i\in I}P_i))$ on $P=\prod_{i\in I} P_i$ equals the product topology $\tau_p$ of the topological spaces $(P_i, \tau_{\textrm{int}}(P_i))$.

Proof. Take a subbasic element of $U\in\tau_i$ and show that it is a member of $\tau_p$. W.l.o.g. we let $U = P\setminus (\uparrow(x_i)_{i\in I})$ where $x_i\in P_i$. Note that $\uparrow(x_i)_{i\in I}$ is a product of closed sets in the spaces $(P_i, \tau_{\textrm{int}}(P_i))$, therefore it is closed in the product topology, so $U\in \tau_p$. Conversely, let $U = \pi_i^{-1}(U_j)$ be subbasic in $\tau_p$ where $\pi_j: P\to P_j$ is the projection map and $U_j = P_j\setminus \uparrow x^*$ for some $x^*\in P_j$. Then $$U = \bigcup \{P\setminus (\uparrow (z_i)_{i\in I}): (z_i)_{i\in I} \in P \text{ and } z_j =x^*\}.$$ So $U\in\tau_i$. QED.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.