How do set theorists view this issue?

266 Views Asked by At

Since the question has changed significantly over the course of last few days, it was suggested to modify it in that light. I have removed those parts which aren't directly related to the main question, as it currently stands now. The current question is mostly just what I wrote in the "Edit:" in the previous version, along with few simplifications and small corrections.

The statement $(**)_1$ was mentioned in the answer (by Noah Schweber). Suppose that a hierarchy such as that in $(**)_1$ exists. Now consider the plausibility argument below of $(**)_1$ being "possibly" true.

Suppose that such a $\omega_1+1$ length hierarchy exists and that the $\omega_1$-point of such a hierarchy is the function $g:\mathbb{N} \rightarrow \mathbb{N}$. Now for any point $\alpha<\omega_1$ we denote the function at the point $\alpha$ as $f_\alpha:\mathbb{N} \rightarrow \mathbb{N}$. Write "$f_1$ eventually dominates $f_2$" as $f_1>f_2$. Now whenever we have $\alpha<\beta<\omega_1$, we also have $f_\alpha < f_\beta$. Furthermore we have $g>f_\alpha$ for all $\alpha<\omega_1$.

Now we want to define a collection of sets $C_i$ (where $i \in \omega$) such that they partition all limit values below $\omega_1$. Meaning all the $C$'s are disjoint and their union is the set of all limit values below $\omega_1$. We define the set $C_n$ (for arbitrary $n$) in the following manner. A given limit element $\alpha<\omega_1$ belongs to $C_n$ if and only if: $$g(n-1) \le f_\alpha(n-1)$$ $$g(x) > f_\alpha(x) \,\,\,\, for \,\, all \,\, x\ge n$$ For the definition of $C_0$ we drop the line $g(n-1) \le f_\alpha(n-1)$ and just consider the second line.

Now we want to consider two separate scenarios.

(1) For all values $i \in \omega$ the least upper bound of values in $C_i$ is strictly less than $\omega_1$. Symbolically we write, for all $i \in \omega$ we have $sup(C_i)<\omega_1$.

In this case loop through limit values from $\omega$ to $\omega_1$ and mark whether they belong to $C_0$ or not. Mark the least upper bound of elements in $C_0$ as $a_0$ (this is first term of sequence). Do this again for $C_1$ and mark the sup of values in $C_1$ as $a_1$. if it turns out that $a_1 \le a_0$ then ignore $a_1$. Otherwise set the second term of sequence to $a_1$. Repeat this all the way $\omega$ number of times.

This gives a sequence for $\omega_1$. This leaves the possibility for scenario (2) which is:

(2) There exist some values $i \in \omega$ such that $sup(C_i)=\omega_1$

Specifically let's just assume that $sup(C_0)=\omega_1$. The definitions below can be easily generalised for any value $k \ne 0$.

Define a function $h^0_\alpha:\mathbb{N} \rightarrow \mathbb{N}$. Denote the smallest element of $C_0$ as $p_0$. We define: $$h^0_\alpha(x)=0 \quad for \,\,\, \alpha \le p_0$$

For $\alpha>p_0$ we have: $h^0_\alpha(x)=max\{f_\beta(x)\,|\, \beta<\alpha$ and $\beta \in C_0 \}$. Now the function $h^0_\alpha$ can only change countable number of times and, furthermore, there must be a point $N_0<\omega_1$ after which it stops changing. So I guess whenever we have $sup(C_0)=\omega_1$ such a value $N_0<\omega_1$ would always exist? If it doesn't exist, then we have a contradiction.

The consideration in the above paragraph applies to all values $k$ for which we have $sup(C_k)=\omega_1$. I also hope, if it is known, someone points to a reference for proof (or an argument) for the existence of $N_k$ whenever $sup(C_k)=\omega_1$.

1

There are 1 best solutions below

20
On BEST ANSWER

What you've written is correct, but it doesn't end in a contradiction.

Here's an example of how $N_0$ can exist - and indeed be small - without really doing anything (I'm going to work with all ordinals, not just limit ordinals, since that's more natural - there's no reason in your current argument to restrict attention to limits):

  • Let $g(x)=x+1$.

  • For each $k\in\omega$, $f_k(x)=\min\{g(x)-1, k\}$.

  • Let $f_\omega=floor(\ln(x+1))$.

Then we have $k\in C_0$ for $k\le\omega$. Moreover, note that the "$h$-sequence" has already stablized by step $\omega$: no matter how we continue the sequence of $f$s, we'll always have $$h_\alpha^0(x)=x$$ for $\alpha\ge\omega$. In your notation, this means $N_0=\omega$.

But this doesn't really do anything to constrain the sequence of $f$s going forward: there is still "lots of room" between $f_\omega$ and $g$. Basically, $\lim_{\alpha\rightarrow\infty,\alpha\in C_0}h_\alpha^0$ doesn't really capture much information.

(Also, your last sentence is incorrect: we could have $C_0$ be everything!)