Adelic Cartier divisors with base conditions and the continuity of volumes (Q824278): Difference between revisions

The paper under review is an extension of Moriwaki's papers [\textit{A. Moriwaki}, J. Algebr. Geom. 18, No. 3, 407--457 (2009; Zbl 1167.14014); Int. Math. Res. Not. 2009, No. 19, 3598--3638 (2009; Zbl 1192.14022); Publ. Res. Inst. Math. Sci. 48, No. 4, 799--898 (2012; Zbl 1281.14017)]. In the above papers, Atsushi Moriwaki introduced the notion of \textit{arithmetic volume function}, which is defined as follows: Definition. \[ \mathrm{vol}(L)=\limsup_{n\rightarrow \infty} \frac{h^{0}(Y, L^{\otimes n)}}{n^{d+1}/(d+1)!} \] By 2011, Huayi Chen proved that the above function is not only continuous, but also differentiable in a certain sense. However, the notion of \(h^{0}(X,L)\) is cumbersome since it treats the archimeadean and non-archimedean places differently. To over-come this issue, it is helpful to come up with a notion of \(\textit{adelic geometry}\) by treating them in the same manner. The author of the present paper tries to overcome this issue by introducing the space of \(l^{1}\)-completed models. First, he the author introduces the notion of adelic \(\mathbb{R}\)-Cartier divisor: \[ (\mathcal{D}, g_{\infty})^{ad}=(D, \sum_{v\in M^{fin}_{K}}g_{v}^{(\mathcal{X},\mathcal{D})}[v]+g_{\infty}[\infty]) \] Then the author introduces the notion of \(l^{1}\) distance defined as \[ d(\mathcal{D}_1, \mathcal{D}_2)=\sum_{v\in M^{fin}_{K}}\sup_{x\in X^{an}_{v}}|g_{v}^{\mathcal{X}_1,\mathcal{D}_1}(x)-g_{v}^{\mathcal{X}_2,\mathcal{D}_2}(x)| \] Unfortunately this notion is not entirely appropriate for Arakelov theory. There are a few reasons: \begin{itemize} \item The space of \(l^1\) functions is too ``rigid'' for analysis purposes at infinity. Since the space of Green functions can be identified as Green operators \(G: H^{s}\rightarrow H^{s+2}\) in a rough sense, the space of \(l^{1}\)-completed functions is ``too small'' for this purpose. \item In general the Green functions will be generalized to Green currents, and the value at a point may not even be well-defined as it is a functional. The issue is avoided in the above definition by using the non-archimedean places only. But then the archimedean place and non-archimedean places will be treated differently. So it defeats the purposes of using adelic language in the first place. \end{itemize} Because of the above difficulties, in the reviewer's opinion, the theorem proved in the paper is not of practical use. It states the following: Theorem. Let \(X\) be a normal, projective, and geometrically connected \(K\)-variety. Let \(V\) be a finite-dimensional \(\mathbb{R}\)-subspace of \(\widetilde{\mathrm{Div}}^{l^1}_{\mathbb{R}}(X)\), let \(|\cdot |_{V}\) be a norm on \(V\), let \(\Sigma\) be a finite set of points on \(X\), and let \(B\in \mathbb{R}_{>0}\). Given any \(\epsilon>0\), there exists a \(\delta>0\) such that \[ |\widehat{\mathrm{vol}}(\overline{D}+(0,f);V)-\widehat{\mathrm{vol}}(\overline{E};V)|\le \epsilon \] for every \(\overline{D}, \overline{E}\in V\) with max \(\{|\overline{D}|_{V}, |\overline{E}_{V}| \}\le \delta\), \(f\in C_{l^1}(X)\) with \(|f|_{l^{1}}\le \delta\), and \(V\in BC_{\mathcal{R}}(X)\) with \(\{c_{X}(v):v(V)>0\} \subset \Sigma\). The author proves a number of interesting results along the way, however. The main theorem above hinges upon the following result: Theorem. Let \(\mathcal{X}\) be a projective arithmetic variety of dimension \(d+1\) over \(\mathrm{Spec}(\mathbb{Z})\). Assume that the generic fibre \(\mathcal{H}_{\mathbb{Q}}\) is smooth over \(\mathrm{Spec}(\mathbb{Q})\). Let \(\overline{\mathcal{L}}=(\overline{\mathcal{L}}_1, \dots, (\overline{\mathcal{L}}_r)\) be a family of \(C^{\infty}\) Hermitian line bundles on \(\mathcal{X}\) and let \(\Sigma\) be a finite set of points on \(\mathcal{X}\). Let \(\mathcal{A}\) be any continuous Hermitian line bundle on \(\mathcal{X}\). There then exista a constant \(C>0\) depending only on \(\mathcal{X}, \overline{\mathcal{L}}, \overline{\mathcal{A}}\) and \(\Sigma\) such that \[ \widehat{l}^{*}(a\cdot \overline{\mathcal{L}}+b\overline{\mathcal{A}}; V)-\widehat{l}^{*}(a\cdot \overline{\mathcal{L}})\le C((|a|_{1}+|b|)^{d}|b|+|a|^{d}_{1}\log|a|_{1}) \] ä It is likely that the above statement will be of use to other researchers in the larger field. The rest of the paper is devoted to set up the framework for \(l^{1}\)-adelic \(\mathbb{R}\)-Cartier divisors. Reviewer's remark: It should be commented that some of the proofs are not explained carefully, for example Proposition 3.13 claimed that the volume for a divisor in the space \(\widetilde{Div}^{l^{1}}_{\mathbb{R},\mathbb{R}}(X)\) is finite, and the proof reduces to the estimation of the volume of a `weaker' divisor \(D_0\). But the estimation for the volume of \(D_0\) has not been proved anywhere (the notion was introduced in page 21), and the proof was suggested to be in the earlier paper Section 2.5 [\textit{H. Ikoma}, Tohoku Math. J. (2) 73, No. 3, 341--401 (2021; Zbl 1493.14043)]. However when we check the earlier paper, no direct corresponding result can be found (the author seems to be referring to Theorem~2.21).