11
$\begingroup$

For a set $X$ we endow the set $\omega^X$ of all functions from $X$ to $\omega$ with the natural partial order $\le$ defined by $f\le g$ iff $f(x)\le g(x)$ for all $x\in X$.

A function $f:\omega^\omega\to\omega^X$ is called monotone if for any $\alpha\le\beta$ in $\omega^\omega$ we have $f(\alpha)\le f(\beta)$ in $\omega^X$.

Question. Is there a monotone function $f:\omega^\omega\to\omega^{\omega_1}$ such that for every countable set $A\subset\omega_1$ and every $\alpha\in\omega^A$ there exists $\beta\in\omega^\omega$ such that $\alpha\le f(\beta){\restriction}_A$?

Remark. By Proposition 2.4.1(2) in this paper, for every monotone function $f:\omega^\omega\to\omega^{\omega_1}$ there exists $\alpha\in\omega^{\omega_1}$ such that for every $\beta\in\omega^\omega$ we have $\alpha\not\le f(\beta)$.

PS. I learned this question from Jerzy Kąkol who arrived to it studying $\mathfrak G$-bases in locally convex spaces.

$\endgroup$

1 Answer 1

5
$\begingroup$

The answer is no, if you require that the monotone function $F \, \colon \, \omega^\omega \rightarrow \omega^{\omega_1}$ is total. I will use fairly standard notation, i.e. $f,g \in \omega^\omega$, $\alpha, \beta \in \omega_1$ and $k,m,n \in \omega$.

The proof works as follows: Towards a contradiction assume that such a function $F$ exists.

First we will construct a sequence $(a_n)_{n \in \omega} \in \omega^\omega$ and find an ordinal $\beta < \omega_1$ such that $$\forall n \in \omega \,\, \exists f \in \prod_{m \leq n} a_m \times \prod_{m > n} \omega \, \colon \, F(f)(\beta) \geq n \, ,$$ where $\prod_{m \leq n} a_m \times \prod_{m > n} \omega =\{ f \in \omega^\omega \, \colon \, \forall m \leq n \,\, f(m) < a_m \}$. The idea behind this is that we can require a bound for finitely many values of an input function $f$ and still make $F(f)(\beta)$ arbitrarily large.

In the second step we construct a sequence $(b_n)_{n \in \omega} \in \omega^\omega$ such that $(b_n)_{n \in \omega} \geq (a_n)_{n \in \omega}$, and a sequence $(f_n)_{n \in \omega}$ such that $$\forall n \in \omega \, \colon \, f_n \in \omega^\omega \, \land \,(b_m)_{m \in \omega} \geq f_n \, \land \, F(f_n)(\beta) \geq n \,.$$ But this leads to a contradiction, since the monotonicity of $F$ implies that $\forall n \in \omega \, \colon \, F((b_m)_{m \in \omega})(\beta) \geq n$.

First step:

We will construct $(a_n)_{n\in \omega}$ by induction. For the base case $n=0$ we claim that $$\exists a_0 \in \omega \,\exists \alpha_0 \in \omega_1 \, \forall A \in [\omega_1 \setminus\alpha_0]^{\leq \aleph_0} \, \colon \, F[\prod_{m = 0} a_0 \times \prod_{m>0} \omega] \,\text{is cofinal in} \, \omega^A\, .$$ This means that already $F\restriction \prod_{m = 0} a_0 \times \prod_{m>0} \omega$ dominates $\omega^A$ for every countable $A \subseteq \omega_1$ above $\alpha_0$.

Towards a contradiction assume that the claim is wrong. Therefore, we can construct a sequence $(A_n)_{n \in \omega}$ such that $A_n \in [\omega_1]^{\leq \aleph_0}$ and $\sup A_n < \min A_{n+1}$, and a sequence of functions $(f_n)_{n \in \omega}$ such that $f_n \in \omega^{A_n}$ and $F\restriction \prod_{m = 0} n \times \prod_{m>0} \omega$ does not dominate $f_n$. But if we define $B:= \bigcup_{n \in \omega} A_n \in [\omega_1]^{\leq \aleph_0}$ and $f:= \bigcup_{n \in \omega} f_n \in \omega^B$, we reach a contradiction, since noting that $\omega^\omega = \bigcup_{n \in \omega} \prod_{m = 0} n \times \prod_{m>0} \omega$, there cannot exist a $g \in \omega^\omega$ such that $F(g) \restriction B \geq f$.

Assume inductively that $a_0,..,a_n$ and increasing $\alpha_0,...,\alpha_n$ have already been defined, such that $\forall A \in [\omega_1 \setminus\alpha_n]^{\leq \aleph_0} \, \colon \, F[\prod_{m \leq n} a_m \times \prod_{m>0} \omega] \,\text{is cofinal in} \, \omega^A$. With a similar argument as before, and again noting that $$\prod_{m \leq n} a_m \times \prod_{m>n} \omega = \bigcup_{k \in \omega} \prod_{m \leq n} a_m \times \prod_{m=n+1} k \times \prod_{m>n+1} \omega$$ we can show that $$\exists a_{n+1} \in \omega \, \exists \alpha_{n+1} \in \omega_1 \setminus \alpha_n \, \forall A \in [\omega_1 \setminus\alpha_{n+1}]^{\leq \aleph_0} \, \colon$$ $$F[\prod_{m \leq n} a_m \times \prod_{m=n+1} a_{n+1} \times \prod_{m>n+1} \omega] \,\text{is cofinal in} \, \omega^A \, .$$ So we can find $a_{n+1}$ and $\alpha_{n+1}$ as required.

Now let $\beta > \sup_{n \in \omega} \alpha_n$. It follows from our construction of the $(a_n)_{n \in \omega}$ that $$\forall n \in \omega \,\, \exists f \in \prod_{m \leq n} a_m \times \prod_{m > n} \omega \, \colon \, F(f)(\beta) \geq n \, .$$

Second step:

Again, we will construct $(b_n)_{n \in \omega}$ and $(f_n)_{n \in \omega}$ by induction. The base case $n=0$ is quite easy: Set $b_0=a_0$ and pick any $f_0 \in \prod_{m = 0} b_0 \times \prod_{m>0} \omega$. Then $F(f_0)(\beta)\geq 0$ trivially holds.

Assume inductively that $b_0,...,b_n$ and $f_0,...,f_n$ have already been constructed such that $$\forall m \leq n \, \colon \, a_m \leq b_m \, \land \, f_m \in \prod_{m \leq n} b_m \times \prod_{m>n} \omega \land \, F(f_m)(\beta) \geq m \, .$$

By using what we have shown in the first step, we can now find $f_{n+1} \in \prod_{m \leq n} b_m \times \prod_{m>n} \omega$ such that $F(f_{n+1})(\beta) \geq n+1$. We set $b_{n+1}:= \max( \max_{m \leq n+1} f_m(n+1), a_{n+1})+1$. It follows that $$\forall m \leq n+1 \, \colon \, a_m \leq b_m \, \land \, f_m \in \prod_{m \leq n+1} b_m \times \prod_{m>n+1} \omega \land \, F(f_m)(\beta) \geq m \, .$$

But now we have reached a contradiction, since $\forall n \in \omega \, \colon \, (b_m)_{m \in \omega} \geq f_n$, and therfore the monotonicity (and totality) of $F$ implies that $\forall n \in \omega \, \colon \, F((b_m)_{m \in \omega})(\beta) \geq n$.

$\endgroup$
5
  • $\begingroup$ I will edit my answer in the evening in order to use better notation and maybe make some points clearer. $\endgroup$ Jun 6, 2020 at 10:26
  • $\begingroup$ So, I will wait for the evening's edition and then will trye to understand your solution. $\endgroup$ Jun 6, 2020 at 13:35
  • $\begingroup$ The proof should now be in a much better and clearer form. Let me now if you have any remarks, or if I've made a mistake. $\endgroup$ Jun 6, 2020 at 19:02
  • $\begingroup$ Now everything seems to be correct and understandable. I am going to write you to an e-mail which I found on the web-page dmg.tuwien.ac.at/fb6/schuerz/home.html Is it ok? $\endgroup$ Jun 6, 2020 at 19:53
  • $\begingroup$ Sure!! I just realized that a comment must have at least 15 characters... $\endgroup$ Jun 6, 2020 at 19:57

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.