Bi-\(\overline{\mathbb{Q}}\)-structures on Hermitian symmetric spaces and quadratic relations between CM periods (Q7285265)

The paper introduces the notion of a bi-\(\bar{\mathbb{Q}}\)-structure. This is used to study quadratic relations between holomorphic periods of CM abelian varieties. The authors define bi-\(\bar{\mathbb{Q}}\)-structures as follows: if \(W_1= 2 \pi i \bar{\mathbb{Q}}\subset \mathbb{C}\) and \(W_2 = \bar{\mathbb{Q}} \subset \mathbb{C}\) are \(\bar{\mathbb{Q}}\) structures on \(\mathbb{C}\), then \(\alpha \in \mathbb{C}\) is called \textit{bi-\(\bar{\mathbb{Q}}\) for \(W_1\)} (resp. \(W_2\)) if \(\alpha \in W_1\) (resp. \(W_2\)) and \(e^{\alpha} \in \bar{\mathbb{Q}}^*\) (resp. \(\alpha = 0\)). The pair \((W_1, W_2)\) is called a \textit{bi-\(\bar{\mathbb{Q}}\)-structure on \(\mathbb{C}\)}. The definition is intended to capture theorems of Gelfond-Schneider and Lindemann which state that if \(\alpha \in \bar{\mathbb{Q}}\) and \(e^{2 \pi i \alpha} \in \bar{\mathbb{Q}}^*\) then \(\alpha \in \mathbb{Q}\), and if \(e^{\alpha} \in \bar{\mathbb{Q}}^*\) then \(\alpha = 0\), respectively. The authors prove that if \(A\) is a \(g\)-dimensional CM abelian variety then the tangent space \(T_0A\) admits a natural bi-\(\bar{\mathbb{Q}}\)-structure, which decomposes in a canonical way as\N\[\NT_0 A = \bigoplus_{j=1}^g \mathbb{C}(\theta_j),\N\]\Nwhere the terms \(\theta_j\) are the \textit{holomorphic periods} of \(A\). Polynomial relations between periods are the source of many open questions. Würtholz's analytic subgroup theorem (WAST) [\textit{G. Wüstholz}, Ann. Math. (2) 129, No. 3, 501--517 (1989; Zbl 0675.10025)] describes linear relations for linear algebraic groups and 1-motives. It implies that holomorphic periods of CM abelian varieties are transcendental, and \(\bar{\mathbb{Q}}\)-linearly independent if \(A\) has no square factors. However, little is known outside of the case of linear relations. The main focus of the paper is to introduce a framework to address the following questions:\N\begin{enumerate}\N\item are all non-trivial \(\bar{\mathbb{Q}}\)-quadratic relations between holomorphic periods \(\bar{\mathbb{Q}}\)-linear combinations of elementary relations, which are relations of the form\N\[\N\theta_j \theta_{j'} - c_{jj' kk'} \theta_k \theta_{k'} = 0\N\]\Nwith \(\{j,j'\} \neq \{k, k'\}\) and \(c_{j j' kk'} \in \bar{\mathbb{Q}}^*\)?\N\item Is\N\[\N\operatorname{dim}_{\bar{\mathbb{Q}}} \bigoplus_{1 \leq j \leq j' \leq g} \bar{\mathbb{Q}} \theta_j \theta_{j} = \frac{g(g+1)}{2}\N\]\Nwhen the Mumford-Tate group of \(A\) is a maximal torus?\N\end{enumerate}\NThe authors introduce a natural bi-\(\bar{\mathbb{Q}}\)-structure on the tangent space at a CM-point of a Shimura variety and prove it has a decomposition much as for \(T_0A\), which they calculate for the moduli of abelian varieties \(\mathbb{A}_g\). They then introduce the `analytic subspace conjecture', which generalises WAST in this setting, and show it provides a positive answer to the aforementioned questions in the case of \(\mathbb{A}_g\).