5
$\begingroup$

There seems to be no publicly available listing of properties of natural (or Hessenberg) addition and multiplication of ordinals. So I'm trying to make one. Please confirm correctness of the following properties or provide corrections, and provide any additional relevant properties.

Recall the definitions: Let $\alpha=\omega^{\alpha_1} + \cdots + \omega^{\alpha_k}$ and $\beta = \omega^{\beta_1}+\cdots+\omega^{\beta_\ell}$ be in CNF (so $\alpha_1\ge \cdots\ge\alpha_k$ and $\beta_1\ge\cdots\ge\beta_\ell$). Let $\gamma_1,\ldots, \gamma_{k+\ell}$ be $\alpha_1,\ldots,\alpha_k,\beta_1,\ldots,\beta_\ell$ sorted in nonincreasing order. Then the natural sum of $\alpha$ and $\beta$ is defined as $\alpha\oplus\beta = \omega^{\gamma_1}+\cdots+\omega^{\gamma_{k+\ell}}$. And the natural product of $\alpha$ and $\beta$ is defined as $\alpha\otimes\beta = \bigoplus_{1\le i\le k, 1\le j\le\ell} \omega^{\alpha_i\oplus\beta_j}$.

Properties

Natural sum and product are commutative and associative. Natural product distributes over natural sum. Meaning:

  • $\alpha\oplus \beta=\beta\oplus\alpha$.
  • $\alpha\oplus(\beta\oplus\gamma)=(\alpha\oplus\beta)\oplus\gamma$.
  • $\alpha\otimes\beta=\beta\otimes\alpha$.
  • $\alpha\otimes(\beta\otimes\gamma)=(\alpha\otimes\beta)\otimes\gamma$.
  • $\alpha\otimes(\beta\oplus\gamma)=\alpha\otimes\beta\oplus\alpha\otimes\gamma$.

Monotonicity:

  • If $\alpha<\beta$ then $\alpha\oplus\gamma<\beta\oplus\gamma$.
  • If $\alpha\le\beta$ then $\alpha\otimes\gamma\le\beta\otimes\gamma$.
  • If $\alpha<\beta$ and $\gamma>0$ then $\alpha\otimes\gamma<\beta\otimes\gamma$.

(For the next properties, recall that if $\alpha=\omega^{\alpha_1}+\cdots+\omega^{\alpha_k}$ is in CNF, then $\alpha\omega = \lim_n\alpha n = \omega^{\alpha_1+1}$, and $\alpha^\omega = \lim_n \alpha^n = \omega^{\alpha_1\omega}$.)

  • $\alpha+\beta\le\alpha\oplus\beta$.
  • $\alpha\beta\le\alpha\otimes\beta$.
  • $\underbrace{\alpha\oplus\cdots\oplus\alpha}_n=\alpha\otimes n$.
  • $\lim_n \alpha\oplus n = \alpha+\omega$. (Not $\alpha\oplus\omega$!)
  • If $\alpha$ and $\beta$ are limit ordinals, then $\alpha\oplus\beta = \lim_{\alpha'<\alpha,\beta'<\beta} \alpha'\oplus\beta'$.
  • $\lim_n \alpha\otimes n = \alpha\omega$. (Not $\alpha\otimes\omega$!)
  • $\lim_n \underbrace{\alpha\otimes\cdots\otimes\alpha}_n = \alpha^\omega$.

Repeated natural product: Recall that every ordinal $\gamma$ can be uniquely decomposed into $\gamma=\omega\beta+n$ for some $\beta$ and some $n$. Then, if we write $\alpha^{[n]}$ to denote $\underbrace{\alpha\otimes\cdots\otimes\alpha}_n$, then it would make sense to write $\alpha^{[\omega\beta+n]}$ to denote $(\alpha^\omega)^\beta\otimes\underbrace{\alpha\otimes\cdots\otimes\alpha}_n$.

Upper bounds:

  • If both $\alpha<\omega^\gamma$ and $\beta<\omega^\gamma$ then $\alpha\oplus\beta<\omega^\gamma$.
  • If both $\alpha<\omega^{\omega^\gamma}$ and $\beta<\omega^{\omega^\gamma}$ then $\alpha\otimes\beta<\omega^{\omega^\gamma}$.
$\endgroup$
7
  • $\begingroup$ en.wikipidia has an article ordinal arithmetic, with some bibliography $\endgroup$ Commented Jan 17 at 9:15
  • $\begingroup$ Right, but that Wikipedia article doesn't even give the usual definition of ordinal product, nevermind properties of the operations! $\endgroup$ Commented Jan 17 at 9:20
  • $\begingroup$ See the answers and comments to the MSE question Ordinal exponentiation identity with natural sum of exponents. In particular, my answer gives some literature references, although unless you're just beginning to look into this topic, I suspect most are things you've already looked at. $\endgroup$ Commented Jan 17 at 16:02
  • $\begingroup$ Hi Dave, I am sure most of these properties appear in different papers. But unfortunately they're not readily accessible to someone who Googles "natural sum and product properties". That's why I think the Internet needs a readily accessible list. The most logical places for such a list would be Wikipedia, or this site. The problem with Wikipedia is that it doesn't allow original work, so I would have to look up each property if it appears somewhere, and then give a reference... $\endgroup$ Commented Jan 17 at 16:21
  • 2
    $\begingroup$ I think your idea is fine, and I was only offering references to possibly verify some of your identities. Also, the references I gave might have additional identities you haven't thought of. And however you decide to (or are able to) put this online for reference purposes, it would be a good idea to include some significant references -- not all of the references I provided would qualify as significant, and there are probably some significant references that I don't know about. $\endgroup$ Commented Jan 17 at 20:19

2 Answers 2

4
$\begingroup$

I covered a bunch of this in my paper Intermediate arithmetic operations on ordinal numbers (arXiv version); see in particular Tables 2 and 3. That's not necessarily everything there is to say on the matter, but I think it covers at least the bulk of what you're asking for!

(Putting this answer up quickly, maybe will come back and edit in more later if needed.)

$\endgroup$
3
  • $\begingroup$ Thanks! So we also have $\alpha^{[\beta\oplus\gamma]} = \alpha^{[\beta]}\otimes\alpha^{[\gamma]}$, using my notation for repeated natural product. $\endgroup$ Commented Jan 19 at 3:52
  • 1
    $\begingroup$ Yeah I guess that's the main addition. I have various further unpublished notes on these operations (well, on how all the operations I discussed relate to one another) but they're kind of a pain to dig through and I'm not sure how much the rest will tell you anything you care about. The stuff I actually published is the better part of it. :P $\endgroup$ Commented Jan 19 at 6:00
  • 1
    $\begingroup$ Well, actually, no, I guess that's not all -- because if you're counting repeated natural multiplication as something worth discussing, then you should also count repeated natural addition, what I called "Jacobsthal multiplication" in my paper. So there are identities about that too, which again you can find in my paper in tables 2 and 3 (they're not due to me mind you but I collected them there because collecting all this together was the point!). $\endgroup$ Commented Jan 19 at 23:59
1
$\begingroup$

Now I see that de Jongh and Parikh also define the "repeated natural product". They do so inductively by letting $\alpha^{[0]}=1$, $\alpha^{[\beta+1]}=\alpha^{[\beta]}\otimes\alpha$, and $\alpha^{[\beta]}=\lim_{\gamma<\beta} \alpha^{[\gamma]}$ for limit $\beta$ (Definition 3.6 in their paper). They prove that $(\omega^{\omega^\alpha})^{[\beta]}=(\omega^{\omega^\alpha})^\beta$ for arbitrary $\beta$, whereas what I wrote in the question implies that $\alpha^{[\beta]}=\alpha^\beta$ for limit $\beta$ and arbitrary $\alpha$.

If I find further properties (e.g. in the literature suggested by Dave L Renfro) I will update this answer.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .