Topological flatness of orthogonal local models in the split, even case. I

Arithmetic aspects of modular and Shimura varieties (11G18) Modular and Shimura varieties (14G35) Simple, semisimple, reductive (super)algebras (17B20)

Abstract: Local models are schemes, defined in terms of linear algebra, that were introduced by Rapoport and Zink to study the 'etale-local structure of integral models of certain PEL Shimura varieties over p-adic fields. A basic requirement for the integral models, or equivalently for the local models, is that they be flat. In the case of local models for even orthogonal groups, Genestier observed that the original definition of the local model does not yield a flat scheme. In a recent article, Pappas and Rapoport introduced a new condition to the moduli problem defining the local model, the so-called spin condition, and conjectured that the resulting "spin" local model is flat. We prove a weak form of their conjecture in the split, Iwahori case, namely that the spin local model is topologically flat. An essential combinatorial ingredient is the equivalence of mu-admissibility and mu-permissibility for two minuscule cocharacters mu in root systems of type D.

Review text

Local models are schemes which are defined in terms of linear algebra that were introduced by Rapoport and Zink to order to study local structures of integral models of certain Shimura varieties over \(p\)-adic fields. A basic requirement for a good integral model, or equivalently for the corresponding local model is flatness. In this paper the author studies the local model attached to a split even orthogonal group \(GO_{2n}\) with the Iwahori level structure. It has been known due to \textit{G. Pappas} [J. Algebr. Geom. 9, No. 3, 577--605 (2000; Zbl 0978.14023)] that the naive local model \(M^{\text{naive}}\) may not be flat in general and a modification is necessary. The desired local model \(M^{\text{loc}}\) may be defined to be the scheme-theoretic closure of the generic fiber in \(M^{\text{naive}}\); however one lacks the good description of this scheme as a functor. \textit{G. Pappas} and \textit{M. Rapoport} provided [J. Algebr. Geom. 12, No. 1, 107--145 (2003; Zbl 1063.14029); Duke Math. J. 127, No. 2, 193--250 (2005; Zbl 1126.14028); J. Inst. Math. Jussieu 8, No. 3, 507--564 (2009; Zbl 1185.14018)] various remedies for the non-flatness of the naive local model \(M^{\text{naive}}\). They defined a closed subscheme \(M^{\text{spin}}\subset M^{\text{naive}}\) by adding a so-called spin condition, and showed evidences of their conjecture: \(M^{\text{spin}}=M^{\text{loc}}\). In the paper under review the author shows that \(M^{\text{spin}}\) is topologically flat, that is, the underlying topological spaces of \(M^{\text{spin}}\) and \(M^{\text{loc}}\) are the same. The method is the same as that in \textit{U. Görtz}'s papers [Adv. Math. 176, No. 1, 89--115 (2003; Zbl 1051.14027); Math. Z. 250, No. 4, 775--790 (2005; Zbl 1085.14022)]. However, in the case of type \(D\), the closed fiber \(M^{\text{spin}}_k\) has two connected components that correspond to two dominant minuscule coweights \(\mu_1\) and \(\mu_2\) of \(GO_{2n}\). The author then carefully shows the equality of \(\mu\)-admissibility and \(\mu\)-permissibility for each \(\mu\in \{\mu_1,\mu_2\}\) which extends work of \textit{R. Kottwitz} and \textit{M. Rapoport} [Manuscr. Math. 102, No.4, 403--428 (2000; Zbl 0981.17003)]. Then it remains to treat a lifting problem for points in maximal cells. This paper is well-written.

Recommendations

Cites work

Cited in

(13)

This page was built for publication: Topological flatness of orthogonal local models in the split, even case. I

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q540428)