0
$\begingroup$

Given $a,b\in\Bbb N$, we can write $a=a_tb^t+a_{t-1}b^{t-1}+\dots+a_1b+a_0$ where $t=\lceil\log_ba\rceil$ and $a_i<b<a$.

(1) Supposing if $b\in\mathcal{O}_K$ where $\mathcal{O}_K$ is ring of integers in number field $K$ with $1<N_{K/\Bbb Q}(b)<a$, then can we expect such a representation of $a\in\Bbb N$?

That is, is $a=a_tb^t+a_{t-1}b^{t-1}+\dots+a_1b+a_0$ where $t=\lceil\log_{N_{K/\Bbb Q}(b)}a\rceil$ and $N_{K/\Bbb Q}(a_i)<N_{K/\Bbb Q}(b)<a$ at every $i\in\{0,\dots,t\}$?

(2) When can we expect such representations of $\Bbb Z$ in general ring of integers $\mathcal{O}_K$ of number fields $K$?

(3) Instead of $[K:\Bbb Q]$ if we take $[K:L]$, then can we use $N_{K/L}$ instead of $N_{K/\Bbb Q}$ in some sense at least if $K,L$ are just real extensions?

(4) Fix $m\in\Bbb N$. Is there subset $S\subseteq\mathcal{O}_K\backslash\Bbb Z$ with $|S|$ where $b\in S\implies N_{K/\Bbb Q}(b)\leq\log m$ so that every $n\in\Bbb N$ with $n<m$ can be written as $$a=a_tb^t+a_{t-1}b^{t-1}+\dots+a_1b+a_0$$ where each $a_i\in\Bbb N$ for some base $b\in S$? That is, is $$\Bbb N_{\leq m}\subseteq\bigcup_{b\in S}\Bbb Z[b]?$$

Can $|S|$ be as small as $\log^c m$ where $c>0$ is fixed?

Comments:

Hurkyl's comment forbids $|S|=1$. So I am guessing $|S|=O(1)$ could be impossible as well.

$(4)$ differs from $(2)$ where $a_i\in\mathcal O_K\backslash\Bbb Z$ is allowed.

$\endgroup$
4
  • $\begingroup$ I would surely call it a restricted form of $b$-adic notation. $\endgroup$
    – user13113
    Commented Aug 30, 2015 at 2:15
  • $\begingroup$ Also, you can't always have $a_i \in \mathbb{N}$: you can have $\mathbb{Z}[b]$ be a proper subset of $\mathcal{O}_K$. I think there are number fields where $\mathbb{Z}[b]$ is a proper subring for every choice of $b$... although the actual theorem I'm remembering is that the span of $\{ 1, b, \ldots, b^{[K:\mathbb{Q}] - } \}$ is a proper subgroup of $\mathcal{O}_K$ for all $b$. $\endgroup$
    – user13113
    Commented Aug 30, 2015 at 2:16
  • $\begingroup$ @Hurkyl That's equivalent. If $b \in \mathcal{O}_K$, then $b$ obeys a monic polynomial with integer coefficients of degree $\leq [K:\mathbb{Q}]$, so the integer span of all powers of $b$ is the same as the first $[K:\mathbb{Q}]$ powers. $\endgroup$ Commented Sep 9, 2015 at 15:59
  • $\begingroup$ You may be interested in the concept of a canonical number system. One reference I know of is pdfs.semanticscholar.org/0ba0/…, but there may be better ones. $\endgroup$ Commented Sep 24, 2018 at 21:41

0

You must log in to answer this question.