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

Let $\mathbb{N}^\mathbb{N}$ denote the set of all functions $f:\mathbb{N}\to\mathbb{N}$ and define a quasi-ordering relation $\leq^*$ on $\mathbb{N}^\mathbb{N}$ by $$f\leq^* g \text{ iff } \exists k\in \mathbb{N} \text{ such that } f(n) \leq g(n) \text{ for all } n\geq k.$$

Let $\tau_i(\mathbb{N}^\mathbb{N})$ be the interval topology of $(\mathbb{N}^\mathbb{N}, \leq^*)$.

Is $(\mathbb{N}^\mathbb{N}, \tau_i(\mathbb{N}^\mathbb{N}))$ connected?

  • $\begingroup$ Does it has the stronger property that any two nonempty open sets has nonempty intersection? $\endgroup$ – Fan Zheng Dec 22 '15 at 20:34
  • $\begingroup$ Good question @FanZheng - I don't know...! $\endgroup$ – Dominic van der Zypen Dec 22 '15 at 20:38
  • $\begingroup$ @FanZheng What sequence could belong both to the open set of sequences which are eventually less than the constant 1 sequence and to the open set of sequences which are eventually greater than the constant 2 sequence? $\endgroup$ – Todd Trimble Dec 22 '15 at 21:22
  • 1
    $\begingroup$ @ToddTrimble If I didn't get it wrong, the open set is defined as the set of sequences not eventually less/greater than a given sequence. So this is a very weak topology, which suggested my question above. $\endgroup$ – Fan Zheng Dec 22 '15 at 21:25
  • $\begingroup$ Read "any of finitely many given sequences" for "a given sequence" above. $\endgroup$ – Fan Zheng Dec 22 '15 at 21:31

I claim that this topology has the stronger property of being hyperconnected, i.e., the intersection of any two nonempty open sets is nonempty.

Indeed, any open set contains a set of the form

$$ U=\bigcap_{i=1}^m \{x:x\not\le^*y_i\}\bigcap\bigcap_{j=1}^n \{x:x\not\ge^*z_j\} $$

so the same can be said of the intersection of two open sets. I claim that

$U$ is nonempty iff every $z_j$ has infinitely many nonzero terms.

It is easy to see this claim implies that the topology is hyperconnected. Now we show the claim.

Clearly this condition is necessary, for if $z_j$ has finitely many nonzero terms, then for every $x\in\mathbb N^{\mathbb N}$ we have $x\ge^*z_j$. Conversely, assume all of $z_j$ has infinitely many nonzero terms, then we can produce an $x\in U$ by the following algorithm:

x=0 //initialize x to the zero array k=0 //k indexes x while true do for i = 1 to m x(k)=y(i)(k)+1 //Repeating this infinitely makes x\not\le^*y_i k=k+1 end for for j = 1 to n while z(j)(k)=0 do k=k+1 end while //This loop must end because z(j) has infinitely many nonzero terms x(k)=0 //Repeating this infinitely makes x\not\ge^*z_j as z(j)(k)>0 end for end while

Edit: As Todd noted in the comment, another natural definition of the interval topology is generated by the open sets $\{x:x^*>y\}$ and $\{x:x^*<z\}$, where $x<^*y$ iff there is $n_0$ such that for all $n>n_0$, $x_n<y_n$.

This topology is very different from the previous one. In fact it is not connected. It can be written as $U\cup V$, where $U$ is the set of all sequences tending to $+\infty$ and $V$ is its complement. We will show that both $U$ and $V$ are open.

Take $x\in U$. Then there is $n_0$ such that for all $n>n_0$, $x_n>0$. Let

$$ y_n=\begin{cases}x_n & n\le n_0\\ x_n-1 & n>n_0 \end{cases}. $$

Then $y\in U$ and $x\in (y,+\infty)\subset U$.

Take $x\in V$. Let $z=x+1$ (termwise). Then $z\in V$ and $x\in (-\infty,z)\subset V$.


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.