A criterion for weak convergence on Berkovich projective space (Q993355)

Let us begin by stating a theorem of Weyl. Let \(a\) be a point in the complex \(N\)-dimensional unit torus over \(\mathbb{C}\) \[ \mathbb{T}^N(\mathbb{C}) = \{(a_{0} : \cdots : a_{N}) \in \mathbb{P}^N(\mathbb{C})\, |\, |a_{0}|=\cdots=|a_{N}|=1\}. \] For any \(\ell\geq 1\), let \(Z_{\ell}\) be the multiset (multiplicities allowed) \(\{a,\ldots,a^\ell\}\). Then the sequence \((Z_{\ell})\) is equidistributed with regard to the Haar measure on the unit torus if and only if \(a\) is non-degenerate in \(\mathbb{G}_{m}^N(\mathbb{C})\) (it does not belong to any smaller algebraic subgroup of \(\mathbb{G}_{m}^N\)). The article under review proves an analogue of this result over complete non-trivially valued non-archimedean fields of residual characteristic 0. If \(K\) is such a field, the unit torus \(\mathbb{T}^N(K)\) is not compact and to get around this problem the author chooses to work with the Berkovich space \({\mathbb{T}}^N_{K}\) associated to it. It is a Hausdorff compact space that contains the latter and carries a reduction map to the algebraic torus \({\mathbb{G}}_{m,\tilde{K}}^N\). The preimage of the generic point by this map consists of one point \(\gamma\), called the Gauss point. We can now state the analogue of Weyl's theorem (Theorem 5). Let \(a\) be a point in the \(N\)-dimensional unit torus over \(K\). For any \(\ell\geq 1\), let \(Z_{\ell}\) be the multiset \(\{a,\ldots,a^\ell\}\). Then the sequence \((Z_{\ell})\) is equidistributed with regard to the Dirac measure \(\delta_{\gamma}\) if and only if \(\tilde{a}\) in non-degenerate in \(\mathbb{G}_{m}^N(\tilde{K})\). On the way to this theorem, the author proves other interesting convergence and equidistribution results. Let us now describe more precisely the contents of the paper. After the first section which is introductory, the second discusses Berkovich affine and projective spaces. The author recalls their definitions and proves their basic properties. The third section contains a convergence result for measures on the Berkovich projective space (Theorem 2). Let \((\mu_{\alpha})\) be a net of unit Borel measures on \({\mathbb{P}}^N_{K}\) (since Berkovich spaces can fail to be metrizable, it is better to consider nets than sequences) and let \(\mu\) be another unit Borel measure on \({\mathbb{P}}^N_{K}\). Then, to prove that \((\mu_{\alpha})\) converges weakly to \(\mu\), it is enough to check it for a special class of functions that come from the absolute values of the evaluations of homogeneous polynomials at the points of \({\mathbb{P}}^N_{K}\). The proof is based on the Stone-Weierstrass theorem. The results of the last two sections are deduced from this convergence theorem. Let \((Z_{\alpha})\) be a net of nonempty finite multisets in \({\mathbb{P}}^N_{K}\). Then Theorem 4 states that \((Z_{\alpha})\) is equidistributed with regard to \(\delta_{\gamma}\) if and only if the reduction net \((\tilde{Z}_{\alpha})\) is weakly generic in \({\mathbb P}^N_{\tilde{K}}\) (which means that \(|\tilde{Z}_{\alpha}\cap W|/|\tilde{Z}_{\alpha}| \to 0\) for all proper Zariski-closed subsets \(W\) of \({\mathbb P}^N_{\tilde{K}}\)). Finally Theorem 5 states the analogue of Weyl's equdistribution theorem. Its proof uses a theorem of Mordell-Lang type for \(\mathbb{G}_{m}^N\) due to Laurent. Let us finally mention that the paper is self-contained and very reader-friendly.