0

Let $\alpha$ be an infinite ordinal. Is there an order-preserving injection from $\langle\alpha\cdot\alpha+\omega,\in\rangle$ to $\langle\mathcal{P}(|\alpha|),\subsetneq\rangle$?

I tried to construct an argument for why such injection cannot exist for the simpler case of $\omega\cdot\omega\to\mathcal{P}(\omega)$, but as I failed doing that, I couldn't formalize a construction that works.

Roy Sht
  • 1,339
  • 1
    Actually, such injections do exist for every infinite $\alpha$. In the specific case of $\alpha=\omega$, we can find something truly impressive: $\mathbb{R}$ (with the usual linear order) embeds into $\mathcal{P}(\omega)$, and so consequently every countable linear order embeds into $\mathcal{P}(\omega)$. This is a standard exercise; for example, see here. More generally, though, you can show by transfinite recursion that whenever $\omega\le\alpha$ and $\beta\le\alpha^+$ there is an embedding of $\beta$ into $\mathcal{P}(\alpha)$. – Noah Schweber Jul 18 '22 at 03:18
  • 3
    Sorry, "$\beta\le\alpha^+$" should of course be "$\beta<\alpha^+$" - there can never be an order-preserving map from $\alpha^+$ to $\mathcal{P}(\alpha)$. – Noah Schweber Jul 18 '22 at 03:45

1 Answers1

0

This is my attempt for a solution:

Lemma 1. If $A$ and $B$ are sets s.t. $|A|\le |B|$, there is an order-preserving injection $\langle \mathcal{P}(A),\subsetneq \rangle \to \langle \mathcal{P}(B),\subsetneq \rangle$.

Proof. Let $f:A\to B$ be injective. Define $h:\mathcal{P}(A)\to\mathcal{P}(B)$ by $h(X)=f[X]$ for all $X\subseteq A$. For $X,Y\in\mathcal{P}(A)$, it is clear that $f[X]\subsetneq f[Y]$ iff $X\subsetneq Y$. $~~~\blacksquare$

Lemma 2. (Without assuming choice!) For an infinite ordinal $\alpha$, we have $|\alpha|=|\alpha\cdot \alpha|=|\alpha\cdot \alpha + \omega|$.

Proof. It is known that for an infinite cardinal $\kappa$ we have $\kappa \cdot \kappa =\kappa =\kappa +\kappa$. Using the (un-recuresive) definitions of ordinal addition and multiplication as the following order types:

$$ \begin{align} \sigma +\tau &:= \operatorname{ort}(\sigma^\frown \tau)&:=\operatorname{ort}(\langle\sigma\sqcup\tau,<_{\text{lex}}\rangle), \\ \sigma \cdot \tau &:= \operatorname{ort}(\sigma\otimes\tau)&:=\operatorname{ort}(\langle\sigma\times\tau,<_{\text{lex}}\rangle), \end{align} $$

We deduce that $|\alpha\cdot\alpha|=|\alpha|\cdot|\alpha|$ and $|\alpha+\omega|=|\alpha|+|\omega|$. Therefore,

$$|\alpha|\le|\alpha+\omega|=|\alpha|+|\omega|\le|\alpha|+|\alpha|=|\alpha|,$$ so $|\alpha|=|\alpha+\omega|$. But $|\alpha\cdot\alpha|=|\alpha|\cdot|\alpha|=|\alpha|$, which finishes the proof. $~~~\blacksquare$

Proposition. For infinite ordinal $\alpha$, there exists an order-preserving injection from $\langle\alpha\cdot\alpha+\omega,\in\rangle$ to $\langle\mathcal{P}(|\alpha|),\subsetneq\rangle$.

Proof. By lemma 1, it's sufficient to find an order-preserving injection from $\alpha\cdot\alpha+\omega$ into $\langle\mathcal{P}(A),\subsetneq\rangle$ for any set $A$ with $|A|=|\alpha|$. From lemma 2, $\gamma=\alpha\cdot\alpha+\omega$ is fitting. From here it's easy - we need to define an order-preserving function $g:\gamma\to \mathcal{P}(\gamma)$, but the identity itself finishes the job, as $\subsetneq$ and $\in$ are exactly the same on an ordinal.$~~~\blacksquare$

Any notes? Is this proof OK?

Roy Sht
  • 1,339