3
$\begingroup$

Let $p$ be a prime, and let $f \in \mathbb{Z}_p[[x_1,\dots,x_d]]$ be a power series convergent on all of $\mathbb{Z}_p^d$. We make the following definition concerning the approximation of $f$ by polynomials:

Definition: Let $U \subset \mathbb{Z}_p^d$ be an open subset. We say that a polynomial $g \in \mathbb{Z}_p[x_1,\dots,z_d]$ is an adequate approximation of $f$ on $U$ if for every $x = (x_1,\dots,x_d) \in U$ there exists a unit $u_x \in 1 + p\mathbb{Z}_p$ such that $f(x) = u_x\times g(x)$. We say that $f$ is adequately approximable if there exists a covering of $\mathbb{Z}_p^d$ by open sets such that $f$ admits an adequate approximation on each.

Question: Let $f$ be as above. Is $f$ adequately approximable?

Progress: I believe that the answer is yes in the single-variable case (i.e., the case where $d = 1$). Indeed, the Weierstrass preparation theorem implies the existence of a polynomial $g(x_1) \in \mathbb{Z}_p[x_1]$ such that $f(x_1)/g(x_1) \in 1 + p\mathbb{Z}_p[[x_1]]$. In other words, in the single-variable case, $f$ admits a single adequate approximation on all of $\mathbb{Z}_p$.

As for the multivariable case, I suspect that the answer is no but am unsure of how to go about finding a counterexample. For instance, take $d = 2$. If $f$ admits a single adequate approximation on all of $\mathbb{Z}_p^2$ by a polynomial $g \in \mathbb{Z}_p[x_1,x_2]$, then $f$ and $g$ share the same vanishing locus, which means that the analytic curve cut out by $f$ is algebraic. I imagine that this is not true of all analytic curves, but the possibility remains that $f$ admits various local adequate approximations that together render $f$ adequately approximable.

I would also note that if $f$ is adequately approximable, then by compactness of $\mathbb{Z}_p^d$, we can approximate $f$ with just finitely many polynomials $g$ over a finite union of open sets.

$\endgroup$

1 Answer 1

6
$\begingroup$

No. Just take $f = x_2 - e^{px_2}$. If $f$ is approximable on some open subset containing a vanishing point of $f$, and thus containing the set of points $(x, e^{px})$ for $x$ in some nonempty open $p$-adic set, then there exists a nonzero polynomial that vanishes on $(x,e^{px})$ for all $x$ in some nonempty open $p$-adic set. But any nonzero polynomial in $x$ and $e^{px}$ is a $p$-adic analytic function that is not identically zero and thus has only finitely many zeroes in any bounded $p$-adic set, which is a contradiction since the open $p$-adic set has infinitely many elements.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.