It is well known that the sum $\alpha+\beta$ of two ordinals $\alpha,\beta$ can be defined "geometrically" as the order type of the sum $(\{0\}\times \alpha)\cup(\{1\}\times\beta)$ endowed with the lexicographic order.

Also the product $\alpha\cdot\beta$ of ordinals $\alpha,\beta$ is the order type of the Cartesian product $\beta\times\alpha$ endowed with the lexicographic order.

What about the exponentiation of ordinals?

Does $\alpha^\beta$ have some nice "geometric'' or combinatorial model?

Maybe as some set of (partial) functions endowed with a suitable well-order?


1 Answer 1


Ordinal exponentiation is a special case of linear order exponentiation. For any linear order $L$, element $a\in L$, and ordinal $\beta$ we can define the $\beta$th power of $L$ at $a$, which I'll call "$L_a^\beta$," as the set of functions $f:\beta\rightarrow L$ such that all but finitely many $\alpha\in\beta$ have $f(\alpha)=a$. The ordering on this set is given by looking at the last point of difference: $$f\trianglelefteq g\iff f=g\mbox{ or } f(\max\{x:f(x)\not=g(x)\})<g(\max\{x:f(x)\not=g(x)\}).$$

For ordinals $\alpha,\beta$ we have $\alpha^\beta=\alpha_0^\beta$. Rosenstein's book treats this in some detail (and is generally an awesome book all-around - it's a huge tragedy that it's so hard to find).

What I like about this definition is that it very nicely complements the definition of cardinal exponentiation: they start with the same basic idea of counting functions between sets, but ordinal exponentiation is the "finite support" version.

  • $\begingroup$ It's not well-ordered. $(1,0,0,...) > (0,1,0,...) > ... > (0,...,0,1,0,...) > ...$. Maybe replace min with max? $\endgroup$ Apr 27, 2020 at 9:02
  • $\begingroup$ @MonroeEskew It would make sense to use $\max$ instead if you want to generalize ordinal exponentiation, but it won't be well-ordered if $L$ isn't anyways. $\endgroup$
    – nombre
    Apr 28, 2020 at 9:27
  • $\begingroup$ @nombre That’s ok... I’m just saying that the above definition doesn’t define $\alpha^\beta$ properly. One needs to use max, by my example. (Isn’t it right? The definition given is the same as the lex order, which corresponds to the rationals for $\alpha=2, \beta=\omega$.) But it is easy to show by induction that max works. $\endgroup$ Apr 28, 2020 at 10:51
  • 1
    $\begingroup$ @MonroeEskew Derp, fixed - thanks! $\endgroup$ Apr 29, 2020 at 13:04
  • 1
    $\begingroup$ Do you need $\beta$ to be well-ordered? It looks to me like this definition still makes sense when $\beta$ is any linear order, so you've really defined a way of exponentiating with two linear orders . . . right? $\endgroup$
    – Will Brian
    Apr 29, 2020 at 13:30

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.