Non-emptiness of Newton strata of Shimura varieties of Hodge type

Dong Uk Lee
DOI: https://doi.org/10.48550/arXiv.1603.04563
2016-03-15
Abstract:For a Shimura variety of Hodge type with hyperspecial level at a prime $p$, the Newton stratification on its special fiber at $p$ is a stratification defined in terms of the isomorphism class of the Dieudonne module of parameterized abelian varieties endowed with a certain fixed set of Frobenius-invariant crystalline cycles ("$F$-isocrystal with $G_{\mathbb{Q}_p}$-structure"). There has been a conjectural group-theoretic description of the F-isocrystals that are expected to show up in the special fiber. We confirm this conjecture by two different methods. More precisely, for any $F$-isocrystal with $G_{\mathbb{Q}_p}$-structure that is expected to appear (in a precise sense), first we construct a special point which has good reduction and whose reduction has associated $F$-isocrystal equal to given one. Secondly, we produce a Kottwtiz triple (with trivial Kottwitz invariant) with the $F$-isocrystal component being the given one. According to a recent result of Kisin which establishes the Langlands-Rapoport conjecture, such Kottwitz triple arises from a point in the reduction.
Number Theory,Algebraic Geometry
What problem does this paper attempt to address?
The problem that this paper attempts to solve is whether the Newton stratification of a specific type (Hodge type) of Shimura variety on the special fiber is empty. Specifically, the author aims to verify a conjectural group - theoretic description, that is, to confirm the existence of all F - isocrystals (F - isocrystal with \( G_{\mathbb{Q}_p} \)-structure) that are expected to appear on the special fiber. ### Core Problems of the Paper 1. **Problem Background**: - For a Shimura variety of Hodge type with superspecial level at a prime \( p \), the Newton stratification on the special fiber at \( p \) is defined by parameterizing the isomorphism classes of Dieudonné modules of abelian varieties. - These abelian varieties are equipped with a set of fixed Frobenius - invariant crystalline periods ("F - isocrystal with \( G_{\mathbb{Q}_p} \)-structure"). 2. **Research Objectives**: - Confirm the existence of all F - isocrystals that are expected to appear on the special fiber. - Specifically, for any F - isocrystal that is expected to appear (in a precise sense), the author uses two different methods to construct special points, which have good reduction, and the associated F - isocrystal after reduction is equal to the given F - isocrystal. 3. **Main Contributions**: - The author proves that the Newton stratification is non - empty by two different methods: - First method: Construct a special point that has good reduction, and the associated F - isocrystal after reduction is equal to the given F - isocrystal. - Second method: Construct a Kottwitz triple (with trivial Kottwitz invariant), whose F - isocrystal part is the given F - isocrystal. ### Main Conclusions According to Kisin's recent results, the Langlands - Rapoport conjecture has been established, and such a Kottwitz triple does come from a point in the special fiber. Therefore, the author confirms the conjecture that the Newton stratification is non - empty. ### Formulas and Symbols To ensure the correctness and readability of formulas, here are some key formulas and symbols involved: - **Dieudonné Module**: Let \( X \) be a \( p \)-divisible group over a perfect field \( k \) of characteristic \( p>0 \), then \( D(X) \) represents the Dieudonné module of \( X \), which provides an equivalent category. \[ D(X)=M \] - **F - Isocrystal**: Let \( k \) be a perfect field of characteristic \( p \), then an F - isocrystal \((M,\Phi)\) is a finite - dimensional \( L(k) \)-vector space \( M \), where \( L(k):=\text{Frac}(W(k)) \), and \(\Phi:M\rightarrow M\) is a \(\sigma\)-linear bijective operator. \[ \Phi:M\rightarrow M,\quad \Phi = \rho_W^\vee(b)(\text{id}_{W^\vee}\otimes\sigma) \] - **Newton Polygon**: Let \(\Phi\) be a \(\sigma\)-linear operator, then the Newton polygon \(\text{NP}(\Phi)\) is a lower - convex (piecewise - linear) polygon determined by the slope sequence of \(\Phi\). \[ \text{NP}(\Phi)=\text{lower - convex polygon} \] By the above methods, the author successfully solves the problem that the Newton stratification is non - empty and provides a theoretical basis for a broader moduli space.