6
$\begingroup$

Let $k$ be a complete non-archimedean field and let $\varphi \colon X \to Y$ be a morphism of rigid analytic spaces over $k$, where $\newcommand{\Sp}{\operatorname{Sp}}Y = \Sp(B)$ is affinoid. Consider the following condition:

$(\dagger)$ The morphism $\varphi$ is separated and there exist two finite admissable affinoid coverings $\mathfrak{U} = (U_i)_{i \in I}$, $\mathfrak{V} = (V_i)_{i \in I}$ of $X$ such that $V_i \Subset_Y U_i$ (i.e. $V_i$ lies relatively compact in $U_i$ w.r.t. $Y$) for all $i \in I$.

A morphism $\varphi \colon X \to Y$ (with $Y$ not necessarily affinoid) is called proper if it is separated and there is an admissable affinoid covering $(W_j)_{j \in J}$ of $Y$ such that for each $j \in J$ the morphism $\varphi^{-1}(W_j) \to W_j$ satisfies condition $(\dagger)$.

If $Y$ is affinoid, then obviously a morphism satisfying condition $(\dagger)$ is proper, whereas the converse is not clear. In fact, Bosch in his book Lectures of Formal and Rigid Geometry writes at the beginning of section 6.4 that $(\dagger)$ is slightly stronger than properness.

Is there an example of a proper morphism with affinoid base which does not satisfy condition $(\dagger)$?

$\endgroup$
3
$\begingroup$

These two notions are actually equivalent (at least if $k$ is the fraction field of a dvr $R$), but I do not know any direct way to see this.

The proof I know heavily uses the theory of formal schemes. The three main results we need are Lemma 2.5, Lemma 2.6 and the statement after Corollary 3.2 from Lutkebohmert's paper ``Formal-algebraic and rigid-analytic geometry''.

Lemma 2.5: Let $\mathfrak W \to \mathfrak V=\text{Spf}(A)$ be a morphism of admissible formal schemes whose generic fibers are affinoid and let $\mathfrak U \subset \mathfrak W$ be an open formal subscheme. Then $\mathfrak U_K$ is relatively compact in $\mathfrak W_K$ over $\mathfrak V_K$ if and only if the Zariski-closure $\overline{\mathfrak U}_0$ of $\mathfrak U_0$ in $\mathfrak W_0$ is proper over $\mathfrak V_0$.

Lemma 2.6: Let $f\colon \mathfrak X \to \mathfrak Y$ be a morphism of admissible formal schemes and let $f_K\colon \mathfrak X_K \to \mathfrak Y_K$ be its generic fiber. If the rigid map $f_K$ is proper, the formal map $f$ is proper.

Claim on page 350: Let $f\colon \mathfrak X \to \mathfrak Y=\text{Spf} B$ be a proper morphism of admissible formal schemes. Let $\mathfrak U$ be a formal open subscheme of $\mathfrak X$ which is affine. Then there exists an admissible blowing-up $\mathfrak X \to \mathfrak X'$ with a center contained in the complement of $\mathfrak U_0$ such that there exists an open subscheme $\mathfrak U'$ of $\mathfrak X'$ whose associated rigid subspace $\mathfrak U'_K$ is affinoid and such that the Zariski-closure $\overline{\mathfrak U_0}$ of $\mathfrak U_0$ in $\mathfrak X'_0$ is contained in $\mathfrak U'_0$.

These results easily imply the desired claim. Indeed, suppose that $g\colon X \to \text{Sp}(A)$ is a proper map. Choose some topologically finitely generated ring of definition $A_0 \subset A$. Then using the standard machinery on formal models, we can find a morphism of admissible formal schemes $f\colon \mathfrak X \to \text{Spf } A_0$ such that the generic fiber $f_K$ is equal to $g$. Then Lemma 2.6 reads that $f$ is proper as a map of formal schemes.

Now comes the key step: we use the last cited claim to find a good covering of $X$. Namely, we choose any covering of $\mathfrak X$ by open affine formal schemes $\mathfrak U_i$. We apply that claim to find an admissible blow-up $\mathfrak X'_i \to \mathfrak X$ with an open formal subscheme $\mathfrak U'_i\subset X'_i$ such that $(\mathfrak U'_i)_K$ is affinoid and $\overline{\mathfrak U_0}$ (that is proper as a closed subscheme of $\mathfrak X'_0$) is contained in $(\mathfrak U'_i)_0$. Lemma 2.5 implies that $(\mathfrak U_i)_K$ is relatively compact in $(\mathfrak U'_i)_K$. The last thing to observe is that $(\mathfrak U_i)_K$ form a finite covering of the rigid space $X$.

P.S. Lutkebohmert's paper is written entirely in the noetherian setup. So if one wants to give a proof along these lines without noetherianness assumption on $\mathcal O_K$, then one needs to reprove these results in the set-up of arbitrary (complete) rank-$1$ valuation rings. My recollection is that it should not be that difficult given the recent finiteness results of Fujiwara and Kato (Fujiwara, Kato ``Foundations of Rigid Geometry I'')

$\endgroup$
1
  • 1
    $\begingroup$ Thank you very much. I already came to suspect this. This should also follow from [Temkin, On local properties of non-archimedean analytic spaces, Cor. 4.5] in complete generality. $\endgroup$ – Jakob Werner Oct 12 '19 at 12:32

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.