1

When looking through the topic here on MSE, one can find in particular the questions P adic numbers number theory and 4-adic numbers and zero divisors which boil down to the facts, that either the inverse limits for the construction of $\mathbb{Z}_2$ and $\mathbb{Z}_4$ agree or one cannot define a suitable norm for composite numbers in parallel with the $p$-adic norms, which would give something new by Ostrowski's Theorem https://proofwiki.org/wiki/Ostrowski%27s_Theorem .

Now the question, have metrics $d_k$ on $\mathbb{Q}$ been considered, which are for $k$ not prime not induced by a norm but which agree with the $p$-adic norm if $k$ is prime? Naively, then for $\mathbb{Q}$ equipped with the metric $d_k$, we could hope that the completion $\bar{\mathbb{Q}}^{d_k}$ gives something that looks like the $p$-adics even for composite numbers. I have no good idea for the definition of $d_k$ but would hope that maybe someone could point me to a reference, if this has already bin considered in the literature.

Jfischer
  • 1,073
  • 9
  • 16
  • One issue to be careful about is that if the metric is not in some sense compatible with the algebraic (field) structure, then the completion has no reason to be a field. I think that is why norms / valuations are defined the way they are, with axioms that demand a very strong compatibility with the additive and multiplicative structure. I'm not saying it might not be possible to get what you want with something that is somewhat more general than the usual $p$-adic or $n$-adic seminorms. – Torsten Schoeneberg Feb 04 '25 at 20:10
  • 1
    Yes, that seems reasonable, the strong requirements are, in my opinion, the reason to extend everything further, since norms are naturally homogenous structures under translation and scaling. But I think continuity of addition and multiplication wrt to the metric should already suffice... – Jfischer Feb 05 '25 at 12:17
  • So it seems. Now I wonder if a metric for which the field operations are continuous can always be "sharpened" (i.e. is in some sense equivalent) to one coming from a (submultiplicative) norm. As in, if one tentatively sets $|x| := d(x,0)$, can one get at least existence of some constant $C$ such that $|ab| \le C|a| |b|, |a+b| \le C\cdot (|a| + |b|)$ or similar. – Torsten Schoeneberg Feb 07 '25 at 05:39

0 Answers0