3. Semipositive metrics [03BE]
Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.
Complete original source context · Original author HTML
3. Semipositive metrics
In this section, is an arbitrary non-archimedean field endow with a non-trivial complete absolute value. We will first introduce semipositive formal metrics. We have seen in Proposition 2.8 that formal metrics are the same as piecewise linear metrics and hence everything applies to piecewise linear metrics as well.
3.1.
Let be a proper scheme over with a line bundle over . We call an algebraic -model of numerically effective (briefly nef) if for every closed curve in which is proper over . Of course, properness implies that is contained in the special fiber . An algebraic metric on is said to be semipositive if there is a nef algebraic -model of such that .
3.2.
The above definition is easily generalized to the analytic setting: Let be a line bundle on a paracompact strictly -analytic variety . A formal -model of is called nef if for any closed curve in the special fiber which is proper over . A formal metric on is called semipositive if there is a nef formal -model of such that .
It will follow from Proposition 3.5 below that we may use any model to test semipositivity of the associated metrics.
Lemma 3.3.
Let be a paracompact strictly -analytic space, a line bundle on and a formal model of . Let be a non-archimedean extension of and the model of obtained by base change. Then is nef if and only if is nef.
Proof.
We remark that . Hence the result follows from the fact that a nef line bundle on a proper variety over remains nef after pull back to . This is proven in the projective case in [EFM, Remark 1.3.25] and the proper case follows from Chow’s lemma and the projection formula. ∎
Lemma 3.4.
Let be a paracompact strictly -analytic space, a line bundle on and a formal model of . Let the model of obtained by putting the reduced structure. Then is nef if and only if is nef.
Proof.
Let be the induced reduced structure on . Since is finite (in fact an immersion), we deduce that the induced map between the special fibers is finite. By projection formula, we conclude that is nef if and only if is nef. ∎
Proposition 3.5.
Let be a -model of . Then is a semipositive formal metric if and only if is a nef formal -model.
Proof.
By definition if is nef, then is semipositive, so we only have to prove the reverse implication. Hence we assume that is a semipositive formal metric and we have to show that is nef. Using Lemma 3.3, we can replace by and hence assume that is algebraically closed.
By definition of semipositivity, there is a nef -model of on some model of with . There exists a model of which dominates both and . Let be the induced morphism. Since the induced morphism on the special fibers is proper and surjective, by the projection formula, is nef if and only is nef. Hence replacing by , we can assume that dominates .
Let be the reduced structure on . Hence is finite. Locally, is given by for some reduced admissible -algebra. Let . It is a strictly -affinoid algebra, and by [BGR84, 6.4.3] is an admissible -algebra, and moreover is finite and induces an isomorphism on the generic fibers. By [BGR84, 7.2.6 Proposition 3], we can glue the morphisms to get a model of such that is finite. In particular, we deduce that the induced morphisms are proper and surjective, and we conclude from the projection formula that is nef if and only if its pull back to is nef.
By construction, is locally of the form , hence we deduce that is locally given by which is reduced. Now we use the fact that on an admissible formal scheme with reduced special fibre and with algebraically closed, the metric determines the model up to isomorphism (see [Gub98, Proposition 7.5]). Using that for the pull-back of to , we deduce that . As above, the pull-back of is nef and hence is nef. ∎
Lemma 3.6.
Let be a proper scheme over , a line bundle on and a model of with . Let be the irreducible components of equipped with their reduced structures. Then is semipositive if and only if for all , is semipositive.
Proof.
For each , let be the closed subscheme of defined as the topological closure of in equipped with the reduced structure. We then get for each a cartesian diagram
Since the morphism is finite surjective, the projection formula shows that is nef on if and only if is nef on for all . ∎
3.7.
Following a suggestion of Tony Yue Yu, we can define semipositivity locally on . We say that a piecewise linear metric on is semipositive in if there is a compact strictly -analytic domain in which is a neighborhood of such that the restriction of to is a semipositive formal metric in the sense of 3.2 (using the equivalence of Proposition 2.8). We say that is semipositive if it is semipositive in all . We will see in Proposition 3.10 that this fits with the definition in 3.2 assuming that is boundaryless.
Definition 3.8.
Let be a piecewise -linear metric on the line bundle over and let . Then is called semipositive in if and only if we may choose a compact strictly -analytic domain which is a neighbourhood of and some integer such that is a semipositive formal metric.
It follows easily from Proposition 3.5 that a piecewise linear metric on is semipositive as a piecewise linear metric if and only if it is semipositive as a piecewise -linear metric.
Proposition 3.9.
Let be a line bundle on a paracompact strictly -analytic space . Let and let be a piecewise -linear metric on .
- (a)
The set of points in where is semipositive is open in .
- (b)
The tensor product of two piecewise -linear metrics which are semipositive in is again semipositive in .
- (c)
Let be a morphism of paracompact strictly -analytic spaces. If is semipositive in , then is semipositive in any point of .
Proof.
In the following result, we need the notion of the boundary of an analytic space as introduced in [Ber90, §2.5, §3.1]. An analytic space without boundary is called boundaryless. Note that the analytification of a scheme locally of finite type over is always boundaryless by [Ber90, Theorem 3.4.1] (boundaryless is called closed there).
Proposition 3.10.
Proof.
The proof follows mainly the arguments in [GK15, Proposition 6.4]. By Lemma 3.3 and Lemma 3.4, we may assume that is algebraically closed and that is reduced. Let be a formal -model of with . We assume that is semipositive in every . We choose a closed curve in which is proper over . We have to show that . By surjectivity of the reduction map , there is such that is the generic point of . Since is semipositive in , there is a compact strictly -affinoid neighborhood of and a nef formal -model of such that over . Using Proposition 3.5, we may always replace the models and by dominating formal -models and the line bundles and by their pull-backs. By [BL93b, Corollary 5.4], we may therefore assume that is a formal open subset of . Then is also a formal -model of and hence Proposition 3.5 shows that is nef.
Since is a neighbourhood of and since is boundaryless, we conclude that the boundary of is the topological boundary of in (see [Ber90, Corollary 2.5.13(ii), Proposition 3.1.3(ii)]). In particular, is no boundary point of as is a neighborhood of . Using [CD12, Lemma 6.5.1], such interior points are characterized by the property that the closure of the reduction in is proper over . We conclude that the closure of in is equal to . Since is nef, it follows that . ∎
Proposition 3.11.
Let and be algebraic metrics of the line bundle over the proper scheme over . Then is an algebraic metric on . If and are semipositive in , then is semipositive in .
Proof.
Since formal and algebraic metrics are the same as noted in Remark 2.5 and hence also the same as piecewise linear metrics, we deduce from Proposition 2.10 (d) that is an algebraic metric. If the given metrics are semipositive in , then it remains to prove that is semipositive in . By base change again, we may assume that is algebraically closed. By Lemma 3.6, we may assume that is a proper variety over .
Let us pick models , and of defining the model metrics , and . There is a -model of on which , and are determined. There is an open neighbourhood of in such that and are semipositive in all points of . We will show that is semipositive in every point of . By [GK15, 6.5], it is equivalent to show that for any closed curve of contained in the reduction of . Moreover, the same result yields that and restrict to nef line bundles on . By [GK15, Theorem 4.1], there is a closed curve in such that is an irreducible component of the special fibre of the closure in . By restriction, we may assume that is a curve and hence is an irreducible component of . Let be the formal completion of and let be the line bundles on induced by the pull-backs of .
We have seen in the proof of Proposition 3.5 that we can associate to a canonical formal model of with reduced special fibre and a canonical finite surjective morphism . So there is a closed curve in which maps onto in . Let be the line bundles on given by pull-back of . Note that are formal models of the metrics on . By projection formula, the line bundles restrict to nef line bundles on and it remains to show that
| (3.11.1) |
Let be the generic point of . Then there is a unique point in with reduction . This follows from [Ber90, Proposition 2.4.4] since has a formal affine open neighbourhood in of the form for a strictly -affinoid algebra . Using , we may assume . Since is algebraic, there is a non-trivial meromorphic section of . Note that the restriction of to the generic fibre induces also a meromorphic section of . The meromorphic section of restricts to the trivial section of and we have
By [Gub98, Proposition 7.5], we deduce that is a global section of . The definition of formal metrics and yield that is the generic fibre of a formal open neighbourhood of . Hence [Gub98, Proposition 7.5] again shows that is a nowhere vanishing regular section of on . We conclude that the restriction of the global section to is not identically zero inducing an effective Cartier divisor on . This shows
Using that is nef on and , we get
proving (3.11.1). ∎