Irreducible components of minuscule affine Deligne-Lusztig varieties (Q1621973)

This paper under review studies the geometry of affine Deligne-Lusztig varieties of hyperspecial level structure in both equi-characteristic and mixed characteristic settings. Let us introduce some notation. \par Let $F$ be a non-Archimedean local field, that is, $F$ is a finite extension of $\mathbb Q_p$ or of the field $\mathbb F_p((t))$ of Laurent series. Denote by $O_F$ and $k_F$ its ring of integers and its residue field, respectively, and fix a uniformizer $\epsilon$. Let $\Gamma$ be the absolute Galois group of $F$. Let $L$ denote the completion of the maximal unramified extension of $F$, $O_L$ its ring of integers, and $k$ the residue field. Denote by $\sigma$ the Frobenius map of $L/F$ and of $k/k_F$. \par Let $G$ be a reductive group scheme over $O_F$ and denote $K=G(O_L)$. Fix $S\subset T\subset B \subset G$, where $S$ is a maximal split torus, $T$ a maximal torus and $B$ a Borel subgroup. For each $\mu\in X_*(T)_{\text{dom}}$ and $b\in G(L)$, one associates the affine Deligne-Lusztig variety \[ X_\mu(b):=\{ gK\in G(L)/K \mid g^{-1} b \sigma(g)\in K \mu(\epsilon) K \}. \] This is a $k$-scheme locally of finite type when $\text{char\,} F=p>0$, and locally of perfectly finite type when $\text{char\,} F=0$ due to \textit{X. Zhu} [Ann. Math. (2) 185, No. 2, 403--492 (2017; Zbl 1390.14072)] and \textit{B. Bhatt} and \textit{P. Scholze} [Invent. Math. 209, No. 2, 329--423 (2017; Zbl 1397.14064)]. Thus, one can study the geometry of affine Deligne-Lusztig varieties even in the mixed characteristic case and non-minuscule coweights $\mu$. It is known that $X_\mu(b)$ is non-empty if and only if the $\sigma$-conjugacy class $[b]$ lies in the Kottwitz set $B(G,\mu)$. \par It was conjectured by Rapoport that $X_\mu(b)$ is equi-dimensional. At that time for the mixed characteristic case, the algebro-geometric structure of $X_\mu(b)$ was endowed by Rapoport-Zink spaces and the conjecture was first restricted to the EL or PEL-cases. In this paper the authors prove this conjecture in its great generality (Theorem 3.4). In particular, it holds true for all equi-characteristic cases and for the mixed characteristic cases where $\mu$ is minuscule, $G$ is classical, $p\neq 2$ and $F$ is unramified over $\mathbb Q_p$. This result is extended from results of \textit{U. Hartl} and \textit{E. Viehmann} [Adv. Math. 229, No. 1, 54--78 (2012; Zbl 1316.14089)], \textit{E. Viehmann} [J. Algebr. Geom. 17, No. 2, 341--374 (2008; Zbl 1144.14040); Doc. Math. 13, 825--852 (2008; Zbl 1162.14033)], \textit{P. Hamacher} [Int. Math. Res. Not. 2015, No. 23, 12804--12839 (2015; Zbl 1349.14156); Math. Z. 287, No. 3--4, 1255--1277 (2017; Zbl 1401.14193)], \textit{M. Chen} et al. [Compos. Math. 151, No. 9, 1697--1762 (2015; Zbl 1334.14017)] and Zhu [loc. cit.]. \par The authors' main result concerns the set $\Sigma(X_\mu(b))$ (resp.~$\Sigma^{\text{top}}(X_\mu(b))$) of (resp. top-dimensional) irreducible components of $X_\mu(b)$. It is known that the set $J_b(F)\backslash \Sigma(X_\mu(b))$ of orbits is finite, where $J_b$ is the twisted centralizer of $b$. For describing the results, we let $\widehat G$ be the dual group of $G$, which is equipped with a maximal split torus $\widehat S$ contained in a maximal $\widehat T$ such that there is a natural identification $X^*(\widehat T)=X_*(T)$. Consider an element $\lambda_G(b)\in X^*(\widehat T^\Gamma)$ which is a lift of the Kottwitz point $k_G(b)\in \pi_1(G)_\Gamma$ such that the restriction $\lambda=\lambda_G(b)|_{\widehat S}\in X^*(\widehat S)=X_*(S)$ is a ``best integral approximation'' of the Newton point $\nu_b\in X_*(S)_{\mathbb Q,\text{dom}}$. The main result (Theorem 5.12) of this paper describes the set $J_b(F)\backslash \Sigma^{\text{top}}(X_\mu(b))$ for minuscule $\mu$, which implies in particular the the following result. \par Theorem 1.4. Let $\mu\in X_*(T)_{\text{dom}}$ be minuscule, $b\in [b]\in B(G,\mu)$, and let $\tilde \lambda$ be a lift of $\lambda_G(b)$. Then there is a canonical surjective map \[ \phi: W\cdot \mu \cap \left (\tilde \lambda+(1-\sigma) X_*(T)\right )\to J_b(F)\ backslash \Sigma^{\text{top}}(X_\mu(b)). \] Moreover, this map is a bijection in the following cases: \par (1) $G$ is split and \par (2) the intersection $[b]\cap \text{Cent}_G(\nu_b)$ is a union of superbasic $\sigma$-conjugacy classes in $\text{Cent}_G(\nu_b)$. \par This gives a vast generalization of previous results of Viehmann [loc. cit.] for the cases $G=\mathrm{GL}_{n,\mathbb Q_p}$ or $G=\mathrm{GSp}_{2n,\mathbb Q_p}$ and of \textit{O. Bültel} and \textit{T. Wedhorn} [J. Inst. Math. Jussieu 5, No. 2, 229--261 (2006; Zbl 1124.11029)] and \textit{I. Vollaard} and \textit{T. Wedhorn} [Invent. Math. 184, No. 3, 591--627 (2011; Zbl 1227.14027)] for the case $G=\mathrm{GU}(1,n-1)_{\mathbb Q_p}$, where the set $J_b(F)\backslash\Sigma(X_\mu(b))$ is a singleton. The authors make a great contribution to the geometry of affine Deligne-Lusztig varieties. These results will be very useful for describing irreducible components of Newton strata of good reduction of more Shimura varieties.