Positivity properties of metrics and delta-forms (Q2001374)

This article deals with potential theory over non-Archimedean fields. In view of applications to Arakelov theory, it is essential to have a positivity notions for forms and currents at one's disposal, and the present text defines and studies such notions. Let \(K\) be an algebraically closed complete non-Archimedean valued field. Recall that \textit{A. Chambert-Loir} and \textit{A. Ducros} have set up a theory of forms and currents on Berkovich \(K\)-analytic varieties in their seminal work [``Formes différentielles réelles et courants sur les espaces de Berkovich'', Preprint, \url{arXiv:1204.6277}]. \textit{W. Gubler} and \textit{K. Künnemann} then extended it to a theory of \(\delta\)-forms and \(\delta\)-currents [Algebra Number Theory 11, No. 1, 77--180 (2017; Zbl 1386.14096)]. In this text, they explore positivity notions in the later setting. The first definition is that of positive forms on tropical cycles (Section 1). Then come positive \(\delta\)-forms on \(K\)-analytic varieties, by pulling back via tropicalisation maps, and positive \(\delta\)-currents by duality (Section 2). It is now possible to define \(\delta\)-plurisubharmonic (\(\delta\)-psh in short) functions: functions \(f\) whose associated \(\delta\)-current \(d'd''f\) is positive, and \(\delta\)-psh metrics on line bundles: metrics \(\| \cdot \|\) such that \(-\log (\|s\|)\) is \(\delta\)-psh for \(s\) a nowhere vanishing section (Section 3). Those notions may not be stable by pull-back under morphisms and the authors define a notion of ``functorial \(\delta\)-psh'' where this additional property is required. The rest of the paper deals with formal metrics on line bundles, which are metrics coming from models (of the space and the line bundle) over the ring of integers \(K^\circ\) (Section 5). A notion of positivity due to S.-W. Zhang, called semi-positivity, is already available is this setting: it is of algebraic nature and corresponds to the requirement that the line bundle be nef on the special fiber of a model. The main theorem of the paper asserts that the following statements are equivalent for the metric: 1) it is semipositive; 2) it is functorial \(\delta\)-psh; 3) it is functorial psh; 4) its first Chern class is a positive \(\delta\)-current; 5) the restriction to any closed curve is positive. In Section 7, the notion of semipositive approximable is compared to the others. One of the main ingredients in the proof of the comparison theorem is a lifting theorem (proved in Section 4) that allows to realize any irreducible closed subset of dimension~\(d\) of the special fiber of a model as an irreducible component of the special fiber of the schematic closure of a closed subscheme of dimension~\(d\) of the model. Finally, in Section 8, the authors consider piecewise smooth metrics (which include formal metrics among others), for which they provide a characterisation of plurisubharmonicity in terms of convexity on tropical charts. They explicitly deal with the examples of canonical metrics on abelian varieties and toric metrics on toric varieties.