2

How do I double the resolution of $2^{\nu_2(x)}$?

$f(x)=2^{\nu_2(x)}$ gives the highest power of $2$ that divides $x$, e.g. $2^{\nu_2(28)}=4$

And this can measure consistently over $\Bbb N, \Bbb Z, \Bbb Z[\frac12], \Bbb Z[\frac16], \Bbb Z_2, \Bbb Q_2$ etc.

As it stands this only measures integer powers of $2$, but I want to "double" its resolution so it measures down to $\sqrt2$ in a consistent way. How do I do this, and what sets does the new function range over consistently?


My thinking is that the answer is to set:

$\displaystyle \large {g(x)=(\sqrt2)^{\nu_2(x^2)}}$

a) does that do it, and

b) what analogues of the above list of sets does this function range over consistently?

In respect of part b), I mean, for example: instead of $\Bbb Z[\frac12]$ it might measure over $\Bbb Z\left[\frac1{\sqrt2}\right]$, and is the completion of $\Bbb Z$ or $\Bbb Q$ with respect to this metric different to $\Bbb Z_2$ and $\Bbb Q_2$?

2 Answers2

5

To quote from an answer I gave to one of your questions six months ago (but with new emphasis added):

As long as a number $\alpha$ is algebraic over $\mathbb{Q}$, there is a natural extension of the $2$-adic absolute value both from $\mathbb{Q}$ to $\mathbb{Q}(\alpha)$ and from $\mathbb{Q}_2$ to $\mathbb{Q}_2(\alpha)$. (EDIT: ... and up to $[\Bbb Q(\alpha):\Bbb Q]$ reasonable extensions of $|\cdot|_2$ to $\Bbb Q(\alpha)$, namely, one for each non-archimedean place of $\Bbb Q(\alpha)$ lying above $2$ ($\Leftrightarrow$ one for each irreducible factor of the $\Bbb Q$-minimal polynomial of $\alpha$ when viewed over $\Bbb Q_2$ $\Leftrightarrow$ one for each distinct prime factor in the prime factorisation of $(2)$ in the ring of integers of $\Bbb Q(\alpha)$. In the example $\alpha=\sqrt{2}$, there is only one such extension of the value, as the prime 2 totally ramifies.)

As in your example, of course the only reasonable extension of the absolute value to $\alpha = \sqrt c$ for $c \in \mathbb{Q}_2$ is $|\sqrt c| := \sqrt{|c|}$. Much more generally, an element $\alpha$ whose minimal polynomial over $\mathbb{Q}_2$ has degree $d$ is given the value $|N_{\mathbb{Q}_2(\alpha)|\mathbb{Q}_2}(\alpha)|^{1/d}$. This is treated extensively in any book on $p$-adics, local fields etc.

With basic knowledge of $p$-adic (multiplicative) absolute values $|\cdot|_p$ and (additive) valuations $v_p$, the last definition reads

$v_2(\alpha) := \frac{1}{d}v_2(N_{\mathbb{Q}_2(\alpha)|\mathbb{Q}_2}(\alpha))$

or in your extremely special case

$v_2(\sqrt{2}) = \frac{1}{2}$

which uniquely defines an extension of $v_2$ to $\Bbb Q(\sqrt2)$ (or rather to the much bigger $\Bbb Q_2(\sqrt2)$), namely, as Somos' answer also states, ,

$$\tilde{\nu}_2(a+b\sqrt{2}) = \min (\nu_2(a), \nu_2(b) + \frac{1}{2}) \qquad \text{for } \qquad a,b \in \Bbb Q \, \text{ (or } \Bbb Q_2 \text{)}.$$

What any good lecture notes, books etc. treat is in what precise sense this extension of the value/valuation is unique, how it leads to the theory of ramified and unramified extensions etc.

On a personal note, I doubt that you followed the lead in the last sentence back then, I doubt you follow it now, and I assume you will ask a similar question in six, twelve, eighteen months again.

  • I'm not sure how those extensions are natural. what would be $|\frac 12(1+ \sqrt {-7})|_2$ ? – mercio Jul 04 '18 at 17:12
  • 1
    Given that $-7$ has a square root in $\Bbb{Q}_2$ I'm not sure what @mercio's point is. Probably related to the fact that $|\frac12(1+\sqrt{-7})|_2$ depends on the choice of the (2-adic) square root. One is $\equiv1\pmod8$ the other $\equiv-1$, so clearly the two choices give different 2-adic values. Torsten did say that $d$ is the degree of the minimal polynomial over $\Bbb{Q}_2$. – Jyrki Lahtonen Jul 04 '18 at 19:27
  • @Mercio: Indeed, that depends on which of the two $\sqrt{-7}$ in $\Bbb Z_2$ you mean. But whichever you want, both (and your expression with them) are elements of $\Bbb Q_2$ (actually $\Bbb Z_2$) already, there is no extension here, $d=1$. Jyrki, I think actually one is $\equiv 3$ and the other $\equiv 5$ mod $8$; mercio's expression, accordingly, has (multiplicative) value 2 or 1, respectively. – Torsten Schoeneberg Jul 04 '18 at 20:02
  • okay for $\Bbb Q_2$ but I still wouldn't say that the extension of the norm from $\Bbb Q$ to $\Bbb Q(\sqrt{-7})$ is natural – mercio Jul 04 '18 at 20:11
  • @TorstenSchoeneberg thank-you. Not sure if personal notes are on topic but I tried David Madore's intro which told me all I already knew then Andrew Baker's "introduction" which quickly ventured into proofs and notation I couldn't follow and consequently learnt little (and 99% of which has nothing to do with my extremely special case). I was nevertheless able to progress to correctly extend this valuation as evidenced by the question, and learn how to find limits of sequences etc. – it's a hire car baby Jul 04 '18 at 21:29
  • ... You've answered for $\Bbb Q(\frac12)$ and $\Bbb Q_2(\frac12)$ but it seems less clear how $\Bbb N, \Bbb Z, \Bbb Z[\frac12], \Bbb Z[\frac16]$ might be extended – it's a hire car baby Jul 04 '18 at 21:29
  • @mercio, I see the point now. One gets potentially different extensions of the value to $\Bbb Q(\alpha)$ for different embeddings $\Bbb Q(\alpha) \hookrightarrow \Bbb Q_2(\alpha)$. But it seems that they agree if the embeddings are identical on the preimage of $\Bbb Q_2$; surely if the mipo of $\alpha$ over $\Bbb Q$ stays irreducible over $\Bbb Q_2$ like in the example $\alpha = \sqrt{2}$. I'll think about this a bit more and try to rephrase. Suggestions welcome. – Torsten Schoeneberg Jul 04 '18 at 21:33
  • 1
    @RobertFrost: What I've actually done is extending the map $v_2: \Bbb Q_2 \rightarrow \Bbb Z$ to a map $\tilde v_2: Q_2(\sqrt{2}) \rightarrow \frac{1}{2}\Bbb Z$. All sets that you write, if they make sense at all, are contained in $\Bbb Q_2(\sqrt{2})$, but they are actually already contained in $\Bbb Q_2$. I thought what you wanted to do was to "measure" also expressions with $\sqrt{2}$ in it. – Torsten Schoeneberg Jul 04 '18 at 21:41
  • 3
    As to the other comment, https://math.stackexchange.com/q/73565/96384 . If you think that you can make any substantial progress with $p$-adic numbers by focussing on something you think is a special case and avoiding to learn at least the basics of the theory (which back then took me, a decent second year university student, a couple of months), you are mistaken. If you could not follow the proofs in one set of notes, try again, try harder, try other resources, or stop pretending to yourself and others that you know anything about $p$-adics. No offense, honest opinion. – Torsten Schoeneberg Jul 04 '18 at 21:56
  • @TorstenSchoeneberg thanks for the advice but the motivation is to find a function $f$ such that $f^2(X)=3x+2^{\nu_2(X)}$ and knowing anything about padics may not be necessary. It is you who said it was a very special case. – it's a hire car baby Jul 05 '18 at 04:24
  • @TorstenSchoeneberg with respect to suggestions relating to mercio's problem, I vaguely remember experiencing some similar situation in which I worked out to always choose the square root which is equivalent to $1\pmod 4$ and not the one $\equiv3\pmod4$. Maybe there's a similar answer here. – it's a hire car baby Jul 05 '18 at 08:26
  • @mercio, I've tried to address the issue in the text, thanks for bringing it up. My number theory has gotten a bit rusty, hope it's correct now. – Torsten Schoeneberg Jul 05 '18 at 23:27
  • Torsten, I was discussing residues modulo eight of the two choices of $\sqrt{-7}$ in $\Bbb{Q}_2$. Sorry about not making that clear. I just forgot to plug them in and calculate the residues of $(1+\sqrt{-7})/2$. One is $\equiv1\pmod4$ the other $\equiv0\pmod4$. The point was that it is not surprising that the $|\cdot|_2$-values of the two choices for $(1\pm\sqrt{-7})/2$ are different because the two numbers are not conjugate over $\Bbb{Q}_2$ in spite of appearances to the contrary. – Jyrki Lahtonen Jul 09 '18 at 05:03
  • @Jyrki, we agree about everything conceptual, but our computations are off. I indeed meant that one of the square roots of $-7$ is $\equiv 3$ and the other $\equiv 5$ mod $8$ (the polynomial $x^2+7$ has two more roots mod $8$, namely $\pm 1$, but I think these do not lift further); accordingly, $(1\pm\sqrt{-7})/2$ is congruent to $2$ resp. $3$ mod $4$. Or not? Maybe I miscalculate. – Torsten Schoeneberg Jul 09 '18 at 20:30
  • No. You're right. I was just carelessly thinking that anything of the form $1+8x$ must have square roots $\equiv\pm1\pmod 8$. The 2-adics are precisely the exception. Sorry about creating confusion. – Jyrki Lahtonen Jul 09 '18 at 20:44
  • Ah, what a relief! I was about to check it with every other method in the nice repertoire https://math.stackexchange.com/q/2298779/96384, now I can lean back and eat cake instead. – Torsten Schoeneberg Jul 09 '18 at 20:59
2

I think what you want already partially exists. Quoting the Wikipedia article p-adic order:

In number theory, for a given prime number $p$, the $p$-adic order or $p$-adic valuation of a non-zero integer $n$ is the highest exponent $\nu$ such that $p^\nu$ divides $n$.

The article goes on to extend the function

The $p$-adic order can be extended into the rational numbers. We can define $\nu_p: \mathbb{Q}\to \mathbb{Z}$ $$ \nu_p(a/b) = \nu_p(a)-\nu_p(b).$$

All you need to do is extend one step further. There is a severe restriction though. You have to define the domain of the function. It should be a field and the extended valuation should have properties that extend those it already has. Given a field, there are field extensions. For example, $\mathbb{Q}(\sqrt{2}).$ You can extend the valuation $\nu_2$ to the extended field. Thus $\, \nu_2(x\sqrt{2}) = \frac12 + \nu_2(x) \,$ if $\, x \in \mathbb{Q}^\times. \,$ Also, if $\, x, \, y \in \mathbb{Q}^\times, \, $ then $\, \nu_2(x + y\sqrt{2}) = \min(\nu_2(x),\nu_2(y\sqrt{2})), \,$ as usual.

Somos
  • 35,251
  • 3
  • 30
  • 76