SOME RESULTS ON AFFINE DELIGNE-LUSZTIG VARIETIES

XUHUA HE

Abstract. The study of affine Deligne-Lusztig varieties originally arose from

arithmetic geometry, but many problems on affine Deligne-Lusztig varieties are
purely Lie-theoretic in nature. This survey deals with recent progress on several
important problems on affine Deligne-Lusztig varieties. The emphasis is on
the Lie-theoretic aspect, while some connections and applications to arithmetic
arXiv:1807.02158v1 [math.AG] 5 Jul 2018

geometry will also be mentioned.

1. Introduction
1.1. Bruhat decomposition and conjugacy classes. Let G be a connected
reductive group over a field k and G = G(k). In this subsection, we assume that
k is algebraically closed. Let B be a Borel subgroup of G and W be the finite Weyl
group of G. The Bruhat decomposition G = ⊔w∈W BwB plays a fundamental role
in Lie theory. This is explained by Lusztig [53] in the memorial conference of
Bruhat:
“By allowing one to reduce many questions about G to questions about the Weyl
group W , Bruhat decomposition is indispensable for the understanding of both the
structure and representations of G.”
Below we mention two examples of the interaction between the Bruhat decom-
position and the (ordinary and twisted) conjugation action of G.
(1) Assume that k = Fq and σ is the Frobenius of k over Fq . We assume that G
is defined over Fq and we denote by σ the corresponding Frobenius morphism
on G. The (classical) Deligne-Lusztig varieties was introduced by Deligne and
Lusztig in their seminal work [8]. For any element w ∈ W , the corresponding
Deligne-Lusztig variety Xw is a subvariety of the flag variety G/B defined by
Xw = {gB ∈ G/B; g −1 σ(g) ∈ BwB}.
By Lang’s theorem, the variety Xw is always nonempty. It is a locally closed,
smooth variety of dimension ℓ(w). The finite reductive group G(Fq ) acts
naturally on Xw and on the cohomology of Xw . The Deligne-Lusztig variety
Xw plays a crucial role in the representation theory of finite reductive groups,
see [8] and [50]. The structure of Xw has also found important applications in
number theory, e.g., in the work of Rapoport, Terstiege and Zhang [63], and
in the work of Li and Zhu [49] on the proof of special cases of the “arithmetic
fundamental lemma” of Zhang [76].
(2) Let k be any algebraically closed field. In a series of papers [52], Lusztig
discovered a deep relation between the unipotent conjugacy classes of G and
2010 Mathematics Subject Classification. 14L05, 20G25.
Key words and phrases. Affine Deligne-Lusztig varieties, loop groups, affine Weyl groups.
X. H. was partially supported by NSF DMS-1463852.
1
2 X. HE

the conjugacy classes of W , via the study of the intersection of the unipotent
conjugacy classes with the Bruhat cells of G.
1.2. Affine Deligne-Lusztig varieties. The main objects of this survey are
affine Deligne-Lusztig varieties, analogous of classical Deligne-Lusztig varieties for
loop groups.
Unless otherwise stated, in the rest of this survey we assume that k = Fq ((ǫ)).
Let σ be the Frobenius morphism of k over Fq ((ǫ)). We assume that G is defined
over Fq ((ǫ)) and we denote by σ the corresponding Frobenius morphism on the
loop group G = G(k). We choose a σ-stable Iwahori subgroup I of G. If G
is unramified, then we also choose a σ-stable hyperspecial parahoric subgroup
K ⊃ I. The affine flag variety F l = G/I and the affine Grassmannian Gr = G/K
(if G is unramified) have natural scheme structures.1
Let S be a maximal k-split torus of G defined over Fq ((ǫ)) and let T be its
centralizer, a maximal torus of G. The Iwahori-Weyl group associated to S is
W̃ = N(k)/T (k)1 ,
where N is the normalizer of S in G and T (k)1 is the maximal open compact
subgroup of T (k). The group W̃ is also a split extension of the relative (finite)
Weyl group W0 by the normal subgroup X∗ (T )Γ0 , where X∗ (T ) is the coweight
lattice of T and Γ0 is the Galois group of k over k (cf. [57, Appendix]). The
group W̃ has a natural quasi-Coxeter structure. We denote by ℓ and 6 the length
function and the Bruhat order on W̃ . We have the following generalization of the
Bruhat decomposition
G = ⊔w∈W̃ IwI,
due to Iwahori and Matsumoto [36] in the split case, and to Bruhat and Tits [2]
in the general case. If G is unramified, then we also have
G = ⊔λ is a dominant coweight Kǫλ K.
Affine Deligne-Lusztig varieties were introduced by Rapoport in [60]. Compared
to the classical Deligne-Lusztig varieties, we need two parameters here: an element
w in the Iwahori-Weyl group W and an element b in the loop group G. The
corresponding affine Deligne-Lusztig variety (in the affine flag variety) is defined
as
Xw (b) = {gI ∈ G/I; g −1bσ(g) ∈ IwI} ⊂ F l.
If G is unramified, one may use a dominant coweight λ instead of an element in
W̃ and define the affine Deligne-Lusztig variety (in the affine Grassmannian) by
Xλ (b) = {gK ∈ G/K; g −1bσ(g) ∈ Kǫλ K} ⊂ Gr.
Affine Deligne-Lusztig varieties are schemes locally of finite type over Fq . Also
the varieties are isomorphic if the element b is replaced by another element b′ in
the same σ-conjugacy class.
A major difference between affine Deligne-Lusztig varieties and classical Deligne-
Lusztig varieties is that affine Deligne-Lusztig varieties have the second parameter:
1One may replace Fq ((ǫ)) by the fraction field of the Witt ring. In that case, the affine
Grassmannian Gr and the affine flag variety F l have the structure of perfect schemes, thanks
to the recent breakthrough of Zhu [79], and of Bhatt and Scholze [1]. Many of the results we
discuss in this survey hold for the fraction field of the Witt ring as well.
 3

the element b, or the σ-conjugacy class [b] in the loop group G; while in the classical
case considered in §1.1, by Lang’s theorem there is only one σ-conjugacy class in
G(Fq ) and thus adding a parameter b ∈ G(Fq ) does not give any new variety.
The second parameter [b] in the affine Deligne-Lusztig varieties makes them
rather challenging to study, both from the Lie-theoretic point of view, and from
the arithmetic-geometric point of view. Below we list some major problems on
the affine Deligne-Lusztig varieties:
• When is an affine Deligne-Lusztig variety nonempty?
• If it is nonempty, what is its dimension?
• What are the connected components?
• Is there a simple geometric structure for certain affine Deligne-Lusztig varieties?
We may also consider the affine Deligne-Lusztig varieties associated to arbitrary
parahoric subgroups, besides hyperspecial subgroups and Iwahori subgroups. This
will be discussed in §7.

1.3. A short overview of X(µ, b). The above questions may also be asked for a
certain union X(µ, b) of affine Deligne-Lusztig varieties in the affine flag variety.
Let µ be a dominant coweight of G with respect to a given Borel subgroup of G
over k (in applications to number theory, µ usually comes from a Shimura datum).
The admissible set Adm(µ) was introduced by Kottwitz and Rapoport in [45]. It
is defined by
Adm(µ) = {w ∈ W̃ ; w 6 tx(µ) for some x ∈ W0 }.
We may explain it in a more Lie-theoretic language. Let GrG be the deforma-
tion from the affine Grassmannian to the affine flag variety [10]. The coherence
conjecture of Pappas and Rapoport [57] implies that the special fiber of the global
Schubert variety Gr G,µ associated to the coweight µ (cf. [78, Definition 3.1]) is
∪w∈Adm(µ) IwI/I. This conjecture was proved by Zhu in [78]. Now we set
X(µ, b) = ∪w∈Adm(µ) Xw (b) ⊂ F l.
This is a closed subscheme of F l and serves as the group-theoretic model for the
Newton stratum corresponding to [b] in the special fiber of a Shimura variety
giving rise to the datum (G, µ).
It is also worth mentioning that, although the admissible set Adm(µ) has a
rather simple definition, it is a very complicated combinatorial object. We refer
to the work of Haines and Ngô [20], and the recent joint work of the author with
Haines [19] for some properties of Adm(µ).

1.4. Current status. Affine Deligne-Lusztig varieties in the affine Grassmannian

are relatively more accessible than the ones in the affine flag variety, mainly due
to the following two reasons:
• The set of dominant coweights is easier to understand than the Iwahori-Weyl
group;
• For Xλ (b) the group G is unramified while for Xw (b), we need to deal with
ramified, or even non quasi-split reductive groups.
4 X. HE

For an unramified group G, we also have the fibration ∪w∈W0 tλ W0 Xw (b) → Xλ (b),
with fibers isomorphic to the flag variety of G(Fq ). Thus much information on
Xλ (b) can be deduced from Xw (b).
Nevertheless, the study of the affine Deligne-Lusztig varieties in affine Grass-
mannian is a very challenging task and has attracted the attention of experts
in arithmetic geometry in the past two decades. It is a major achievement in
arithmetic geometry to obtain a fairly good understanding on these varieties.
As to the affine Deligne-Lusztig varieties in the affine flag varieties, the situation
is even more intriguing. We have made significant progress in the past 10 years in
this direction, yet many aspects of Xw (b) remain rather mysterious. I hope that
by combining various Lie-theoretic methods together with arithmetic-geometric
methods, our knowledge on affine Deligne-Lusztig varieties will be considerably
advanced.
In the rest of the survey, we will report on some recent progress on the affine
Deligne-Lusztig varieties.

Acknowledgement. We thank Ulrich Görtz, Urs Hartl, George Lusztig, Michael

Rapoport, Sian Nie and Rong Zhou for useful comments.

2. Some relation with affine Hecke algebras

2.1. The set B(G) and Kottwitz’s classification. Let B(G) be the set of σ-
conjugacy classes of G. Kottwitz [41] and [42] gave a classification of the set B(G),
generalizing the Dieudonné-Manin classification of isocrystals by their Newton
polygons. Any σ-conjugacy class [b] is determined by two invariants:
• The element κ([b]) ∈ π1 (G)Γ , where Γ is the Galois group of k over Fq ((ǫ));
• The Newton point νb in the dominant chamber of X∗ (T )Γ0 ⊗ Q.
A different point of view, which is quite useful in this survey, is the relation
between the set B(G) with the set B(W̃ , σ) of σ-conjugacy classes of W̃ . Recall
that W̃ = N(k)/T (k)1 . The natural embedding N(k) → G induces a natural
map Ψ : B(W̃ , σ) → B(G). By [14] and [26], the map Ψ is surjective. The map Ψ
is not injective. However, there exists an important family B(W̃ , σ)str of straight
σ-conjugacy classes of W̃ . By definition, a σ-conjugacy class O of W̃ is straight
if it contains an element w ∈ O such that ℓ(wσ(w) · · · σ n−1 (w)) = nℓ(w) for all
n ∈ N. The following result is discovered in [26, Theorem 3.7].
Theorem 2.1. The map Ψ : B(W̃ , σ) → B(G) induces a bijection
B(W̃ , σ)str ←→ B(G).
This result gives the parametrization of the σ-conjugacy classes of G in terms of
the set of straight σ-conjugacy classes of its Iwahori-Weyl group W̃ . In particular,
the two parameters occurring in the definition of the affine Deligne-Lusztig variety
Xw (b) are all from W̃ .
Note that the affine Deligne-Lusztig variety Xw (b) is closely related to the in-
tersection IwI ∩ [b]. This intersection is very complicated in general. However, it
is discovered in [26] that for certain elements w ∈ W̃ , the intersection IwI ∩ [b]
 5

equals IwI. More precisely, we denote by W̃σ−min the set of elements in W̃ that
are of minimal length in their σ-conjugacy classes. Then
For w ∈ W̃σ−min , IwI ⊂ [b] if [b] = Ψ(w).
This serves as the starting point of the reduction method for affine Deligne-Lusztig
varieties Xw (b) for arbitrary w.
2.2. “Dimension=Degree” theorem. Deligne and Lusztig introduced in [8] a
reduction method to study the classical Deligne-Lusztig varieties. Their method
works for the affine Deligne-Lusztig varieties as well. Some combinatorial proper-
ties of affine Weyl groups established in joint work with Nie [29] allow us to reduce
the study of Xw (b) for any w, via the reduction method à la Deligne and Lusztig,
to the study of Xw (b) for w ∈ W̃σ−min .
The explicit reduction procedure, however, is rather difficult to keep track of. In
[26], we discovered that the same reduction procedure appears in a totally different
context as follows.
Let H be the affine Hecke algebra (over Z[v ±1 ]) associated to W̃ . Let [H̃, H̃]σ be
the σ-twisted commutator, i.e. the Z[v ±1 ]-submodule of H generated by [h, h′ ]σ =
hh′ − h′ σ(h). By [29], the σ-twisted cocenter H = H/[H, H]σ has a standard basis
given by {TO }, where O runs over all the σ-conjugacy classes of W̃ . Thus for any
w ∈ W̃ , we have X
Tw ≡ fw,O TO mod [H, H]σ .
O
The coefficients fw,O ∈ N[v − v −1 ], which we call the class polynomials (over
v − v −1 ). We have the following “dimension=degree” theorem established in [26].
Theorem 2.2. Let b ∈ G and w ∈ W̃ . Then
1

dim(Xw (b)) = max ℓ(w) + ℓ(O) + deg(fw,O ) − hνb , 2ρi.
O;Ψ(O)=[b] 2

Here ℓ(O) is the length of any minimal length element in O and ρ is the half
sum of positive roots in G. Here we use the convention that the dimension of an
empty variety and the degree of a zero polynomial are both −∞. Thus the above
theorem reduces the nonemptiness question and the dimension formula of Xw (b)
to some questions on the class polynomials fw,O for Ψ(O) = [b].
The explicit computation of the class polynomials is very difficult at present.
Note that there is a close relation between the cocenter and representations of
affine Hecke algebras [7]. One may hope that some progress in the representa-
tion theory of affine Hecke algebras would also advance our knowledge on affine
Deligne-Lusztig varieties. At present, we combine the “dimension=degree” theo-
rem together with some Lie-theoretic techniques, and the results on Xλ (b) in the
affine Grassmannian established previously by arithmetic-geometric method, to
obtain some explicit answers to certain questions on Xw (b) and on X(µ, b).

3. Nonemptiness pattern
3.1. Mazur’s inequality. In this subsection, we discuss the non-emptiness pat-
terns of affine Deligne-Lusztig varieties. Here Mazur’s inequality plays a crucial
role.
6 X. HE

In [54], Mazur proved that the Hodge slope of any F -crystal is always larger
than or equal to the Newton slope of associated isocrystal. The converse was
obtained by Kottwitz and Rapoport in [46]. Here we regard the Newton slope
and Hodge slope as elements in Qn+ = {a1 , · · · , an ; a1 > · · · > an } and the partial
order in Qn+ is the dominance order, i.e. (a1 , · · · , an )  (b1 , · · · , bn ) if and only if
a1 6 b1 , a1 + a2 6 b1 + b2 , · · · , a1 + · · · + an−1 6 b1 + · · · + bn−1 , a1 + · · · + an =
b1 + · · · + bn .
Note that Qn+ is the set of rational dominant coweights for GLn . The dominant
order can be defined for the set of rational dominant coweights for any reductive
group. This is what we use to describe the nonemptiness pattern of some affine
Deligne-Lusztig varieties.

3.2. In the affine Grassmannian. For Xλ (b) in the affine Grassmannian, we

have a complete answer to the nonemptiness question.
Theorem 3.1. Let λ be a dominant coweight and b ∈ G. Then Xλ (b) 6= ∅ if and
only if κ([b]) = κ(λ) and νb  λ.
The “only if” part was proved by Rapoport and Richartz in [61], and by Kot-
twitz in [43]. The “if” part was proved by Gashi [11]. The result also holds if the
hyperspecial subgroup of an unramified group is replaced by a maximal special
parahoric subgroup of an arbitrary reductive group. This was obtained in [26]
using the “dimension=degree” Theorem 2.2.

3.3. In the affine flag. Now we consider the variety Xw (b) in the affine flag
variety.
(i) We first discuss the case where [b] is basic, i.e., the corresponding Newton
point νb is central in G (and thus Mazur’s inequality is automatically satisfied).

Theorem 3.2. Let G be a quasi-split group. Let [b] ∈ B(G) be basic and w ∈ W̃ .
Then Xw (b) 6= ∅ if and only if there is no “Levi obstruction”.
The “Levi obstruction” is defined in terms of the P -alcove elements, introduced
by Görtz, Haines, Kottwitz, and Reuman in [14]. The explicit definition is techni-
cal and we omit it here. This result was conjectured by Görtz, Haines, Kottwitz,
and Reuman in [14] for split groups and was established in joint work with Görtz
and Nie [16] for any quasi-split group. Note that the “quasi-split” assumption
here is not essential as one may relate Xw (b) for any reductive group G to another
affine Deligne-Lusztig variety for the quasi-split inner form of G. We refer to [28,
Theorem 2.27] for the explicit statement in the general setting.
(ii) For any nonbasic σ-conjugacy class [b], one may ask for analogues of “Mazur’s
inequality” and/or the “Levi obstruction” in order to describe the nonemptiness
pattern of Xw (b). This is one of the major open problems in this area. We refer to
[14, Remark 12.1.3] for some discussion in this direction. As a first step, one may
consider the conjecture of Görtz-Haines-Kottwitz-Reumann [14, Conjecture 9.5.1
(b)] on the asymptotic behavior of Xw (b) for nonbasic [b]. Some affirmative answer
to this conjecturewas given in [28, Theorem 2.28] in the case where [b] = [ǫλ ] for
some dominant coweight λ.
 7

3.4. Kottwitz-Rapoport conjecture. To describe the nonemptiness pattern on

the union X(µ, b) of affine Deligne-Lusztig varieties in the affine flag variety, we
recall the definition of neutrally acceptable σ-conjugacy classes introduced by Kot-
twitz in [42],
B(G, µ) = {[b] ∈ B(G); κ([b]) = κ(µ), νb 6 µ⋄ },
where µ⋄ is the Galois average of µ.
By Theorem 3.1, Xµ (b) 6= ∅ if and only if [b] ∈ B(G, µ). We have a similar result
for the union X(µ, b) of affine Deligne-Lusztig varieties in the affine flag variety.
Theorem 3.3. Let [b] ∈ B(G). Then X(µ, b) 6= ∅ if and only if [b] ∈ B(G, µ).
This result was conjectured by Kottwitz and Rapoport in [46] and [60]. The
“only if” part is a group-theoretic version of Mazur’s inequality and was proved by
Rapoport and Richartz for unramified groups in [61, Theorem 4.2]. The “if” part
is the “converse to Mazur’s inequality” and was proved by Wintenberger in [75]
for quasi-split groups. The general case in both directions was established in [27]
by a different approach, via a detailed analysis of the map Ψ : B(W̃ ) → B(G), of
the partial orders on B(G) (an analogy of Grothendieck’s conjecture for the loop
groups) and of the maximal elements in B(G, µ) [30].
As we mentioned in §3.3, for a single affine Deligne-Lusztig variety Xw (b), one
may reduce the case of a general group to the quasi-split case. However, for the
union of affine Deligne-Lusztig varieties, the situation is different. There is no
relation between the admissible set Adm(µ) (and hence X(µ, b)) for an arbitrary
reductive group and its quasi-split inner form. This adds essential difficulties in
the study of X(µ, b) for non quasi-split groups.
Rad and Hartl in [59] established the analogue of the Langlands-Rapoport con-
jecture [48] for the rational points in the moduli stacks of global G-shtukas, for ar-
bitrary connected reductive groups and arbitrary parahoric level structure. They
described the rational points as a disjoint union over isogeny classes of global
G-Shtukas, and then used Theorem 3.3 to determine which isogeny classes are
nonempty.

4. Dimension formula
4.1. In the affine Grassmannian. For Xλ (b) in the affine Grassmannian, we
have an explicit dimension formula.
Theorem 4.1. Let λ be a dominant coweight and b ∈ G. If Xλ (b) 6= ∅, then
1
dim Xλ (b) = hλ − νb , ρi − defG (b),
2
where defG (b) is the defect of b.
The dimension formula of Xλ (b) was conjectured by Rapoport in [60], inspired
by Chai’s work [3]. The current reformulation is due to Kottwitz [44]. For split
groups, the conjectural formula was obtained by Görtz, Haines, Kottwitz and
Reuman [13] and Viehmann [68]. The conjectural formula for general quasi-split
unramified groups was obtained independently by Zhu [79] and Hamacher [21].
8 X. HE

4.2. In the affine flag variety. Now we consider Xw (b) in the affine flag variety.
Theorem 4.2. Let [b] ∈ B(G) be basic and w ∈ W̃ be an element in the shrunken
Weyl chamber (i.e., the lowest two-sided cell of W̃ ). If Xw (b) 6= ∅, then
1
dim Xw (b) = (ℓ(w) + ℓ(ησ (w)) − defG (b)).
2
Here ησ : W̃ → W0 is defined in [14].
This dimension formula was conjectured by Görtz, Haines, Kottwitz, and Reuman
in [14] for split groups and was established for residually split groups in [26]. The
proof in [26] is based on the “dimension=degree” Theorem 2.2, some results on
the σ-twisted cocenter H of affine Hecke algebra H, together with the dimension
formula of Xλ (b) (which was only known for split groups at that time). The di-
mension formula for arbitrary reductive groups (under the same assumption on b
and w) is obtained by the same argument in [26], once the dimension formula of
Xλ (b) for quasi-split unramified groups became available, cf. Theorem 4.1.
Note that the assumption that w is contained in the lowest two-sided cell is an
essential assumption here. A major open problem is to understand the dimension
of Xw (b) for [b] basic, when w is in the critical stripes (i.e., outside the lowest
two-sided cell). So far, no conjectural dimension formula has been formulated.
However, the “dimension=degree” Theorem 2.2 and the explicit computation in
low rank cases [14] indicate that this problem might be closely related to the theory
of Kazhdan-Lusztig cells. I expect that further progress on the affine cellularity of
affine Hecke algebras, which is a big open problem in representation theory, might
shed new light on the study of dim Xw (b).
I also would like to point out that affine Deligne-Lusztig varieties in affine Grass-
mannians are equi-dimensional, while in general affine Deligne-Lusztig varieties in
the affine flag varieties are not equi-dimensional.
4.3. Certain unions. We will see in §8 that for certain pairs (G, µ), X(µ, b)
admits some simple geometric structure. In these cases, one may write down an
explicit dimension formula for X(µ, b). Outide these case, very little is known for
dim X(µ, b).
Here we mention one difficult case: the Siegel modular variety case. Here G =
Sp2g and µ is the minuscule coweight. It was studied by Görtz and Yu in [18], in
2
which they showed that for basic [b], dim X(µ, b) = g2 if g is even and g(g−1)
2
6
g2
dim X(µ, b) 6 [ 2 ] if g is odd. It would be interesting to determine the exact
dimension when g is odd.
5. Hodge-Newton decomposition
To study the set-theoretic and geometric properties of affine Deligne-Lusztig
varieties, a very useful tool is to reduce the study of affine Deligne-Lusztig varieties
of a connected reductive group to certain affine Deligne-Lusztig varieties of its
Levi subgroups. Such reduction is achieved by the Hodge-Newton decomposition,
which originated in Katz’s work [37] on F -crystals with additional structures. In
this section, we discuss its variation for affine Deligne-Lusztig varieties in affine
Grassmannians, and further development on affine Deligne-Lusztig varieties in
affine flag varieties, and on the union of affine Deligne-Lusztig varieties.
 9

5.1. In the affine Grassmannian. For affine Deligne-Lusztig varieties in the

affine Grassmannian, Kottwitz in [43] (see also [69]) established the following
Hodge-Newton decomposition, which is the group-theoretic generalization of Katz’s
result. Here the pair (λ, b) is called Hodge-Newton decomposable with respect to
a proper Levi subgroup M if b ∈ M and λ and b have the same image under the
Kottwitz’s map κM for M.
Theorem 5.1. Let M be a Levi subgroup of G and (λ, b) be Hodge-Newton de-
composable with respect to M. Then the natural map XλM (b) → XλG (b) is an
isomorphism.
5.2. In the affine flag variety. For affine Deligne-Lusztig varieties in affine flag
varieties, the situation is more complicated, as the Hodge-Newton decomposability
condition on the pairs (w, b) is rather difficult. As pointed out in [14], “It is striking
that the notion of P -alcove, discovered in the attempt to understand the entire
emptiness pattern for the Xx (b) when b is basic, is also precisely the notion needed
for our Hodge-Newton decomposition.”
The Hodge-Newton decomposition for Xw (b) was established by Görtz, Haines,
Kottwitz and Reuman in [14].
Theorem 5.2. Suppose that P = MN is a semistandard Levi subgroup of G and
w ∈ W̃ is a P -alcove element in the sense of [14]. Let b ∈ M. Then the natural
map XwM (b) → XwG (b) induces a bijection
J M \X M (b) ∼
b w = J G \X G (b).
b w

5.3. Certain unions. For X(µ, b), the Hodge-Newton decomposability condition
is still defined on the pair (µ, b). However, the precise condition is more compli-
cated than in §5.1 as we consider arbitrary connected reductive groups, not only
the unramified ones. We refer to [17, Definition 2.1] for the precise definition. The
following Hodge-Newton decomposition for X(µ, b) was established in a joint work
with Görtz and Nie [17].
Theorem 5.3. Suppose that (µ, b) is Hodge-Newton decomposable with respect to
some proper Levi subgroup. Then
X(µ, b) ∼
G ′
= X M (µP ′ , bP ′ ),
P ′ =M ′ N ′
′
where P runs through a certain finite set of semistandard parabolic subgroups.
The subsets in the union are open and closed.
We refer to [17, Theorem 3.16] for the precise statement. Note that an essential
new feature is that unlike the Hodge-Newton decomposition of a single affine
Deligne-Lusztig variety (e.g. Xλ (b) or Xw (b)) where only one Levi subgroup is
involved, in the Hodge-Newton decomposition of X(µ, b) several Levi subgroups
are involved.
Thus, the statement here is more complicated than the Hodge-Newton decom-
position of Xλ (b) and Xw (b). But this is consistent with the fact that the Newton
strata in the special fiber of Shimura varieties with Iwahori level structure are
more complicated than those with hyperspecial level structure. I believe that the
Hodge-Newton decomposition here would help us to overcome some of the dif-
ficulties occurring in the study of Shimura varieties with Iwahori level structure
10 X. HE

(as well as arbitrary parahoric level structures). We will see some results in this
direction in §6 and in §8.

6. Connected components
In this subsection, we discuss the set of connected components of some closed
affine Deligne-Lusztig varieties, e.g.
Xλ (b) := ∪λ′ λ Xλ′ (b) and X(µ, b) = ∪w∈Adm(µ) Xw (b).
The explicit description of the set of connected components has some important
applications in number theory, which we will mention later.
Note that affine Grassmannians and affine flag varieties are not connected in
general, and their connected components are indexed by π1 (G)Γ0 . This gives the
first obstruction to the connectedness. The second obstruction comes from the
Hodge-Newton decomposition, which we discussed in §5. One may expect that
these are the only obstructions. We have the following results.
Theorem 6.1. Assume that G is an unramified simple group and that (λ, b) is
Hodge-Newton indecomposable. Then
π0 (Xλ (b)) ∼
= π1 (G)σ .Γ0

This was first proved by Viehmann for split groups, and then by Chen, Kisin
and Viehmann [5] for quasi-split unramified groups and for λ minuscule. The
description of π0 (Xλ (b)) for G quasi-split unramified, and λ non-minuscule, was
conjectured in [5] and was established by Nie [56].
Note that the minuscule coweight case is especially important for applications
in number theory. Kisin [38] proved the Langlands-Rapoport conjecture for mod-
p points on Shimura varieties of abelian type with hyperspecial level structure.
Compared to the function field analogous of Langlands-Rapoport conjecture [59],
there are extra complication coming from algebraic geometry and the explicit
description of the connected components of X(µ, b) in [5] is used in an essential
way to overcome the complication.
Theorem 6.2. Let µ be a dominant coweight and b ∈ G. Assume that [b] ∈
B(G, µ) and that (µ, b) is Hodge-Newton indecomposable. Then
(1) If [b] is basic, then π0 (X(µ, b)) ∼
= π1 (G)σΓ0 .
∼
(2) If G is split, then π0 (X(µ, b)) = π1 (G).
Here part (1) was obtained in joint work with Zhou [33]. As an application, we
verified the Axioms in [32] for certain PEL type Shimura varieties. In [33], the set
of connected components of X(µ, b) was also studied for nonbasic b. We proved the
in a residually split group, the set of connected components is “controlled” by the
set of straight elements, together with the obstruction from the corresponding Levi
subgroup. Combined with the work of Zhou [77], we verified in the residually split
case, the description of the mod-p isogeny classes on Shimura varieties conjectured
by Langlands and Rapoport [48]. Part (2) is recent work of Chen and Nie [6].
We would like to point out that in the statement, the following two conditions
are essential:
• The σ-conjugacy class [b] is neutrally acceptable, i.e. [b] ∈ B(G, µ). This
condition comes from the Kottwitz-Rapoport conjecture (see Theorem 3.3).
 11

• The pair (µ, b) is Hodge-Newton indecomposable. In the general case, we need to

apply the Hodge-Newton decomposition (see Theorem 5.3). As a consequence,
several π1 (M) are involved in the description of π0 (X(µ, b)) in general.

7. Arbitrary parahoric level structure

7.1. Parahoric level versus Iwahori level. Let K ′ ⊃ I be a standard parahoric
subgroup of G and WK ′ be the finite Weyl group of K ′ . We define
X(µ, b)K ′ = {gK ′ ∈ G/K ′ ; g −1bσ(g) ∈ K ′ Adm(µ)K ′ }.
If K ′ = I, then X(µ, b)K ′ = X(µ, b). If G is unramified, µ is minuscule and
K ′ = K is a hyperspecial parahoric subgroup, then X(µ, b)K ′ = Xµ (b). As we have
mentioned, the varieties Xµ (b) (resp. X(µ, b)) serve as group-theoretic models for
the Newton strata in the special fiber of Shimura varieties with hyperspecial (resp.
Iwahori) level structure. The variety X(µ, b)K ′ plays the same role in the study of
Shimura varieties with arbitrary parahoric level structure.
The following result relates X(µ, b)K ′ for an arbitrary parahoric subgroup K ′
with X(µ, b) (for the Iwahori subgroup I).
Theorem 7.1. The projection map G/I → G/K ′ induces a surjection
X(µ, b) ։ X(µ, b)K ′ .
This was conjectured by Kottwitz and Rapoport in [47] and [60] and was proved
in [27]. This fact allows one to reduce many questions (e.g. nonemptiness pattern,
connected components, etc.) of X(µ, b)K ′ for arbitrary K ′ to the same questions
for X(µ, b). In fact, the statements in Theorem 3.3 and Theorem 6.2 hold if
X(µ, b) is replaced by X(µ, b)K ′ for an arbitrary parahoric subgroup K ′ .

7.2. Lusztig’s G-stable pieces. I would like to draw attention to some crucial
ingredient in the proof, which has important applications in arithmetic geometry.
Note that I Adm(µ)I $ K ′ Adm(µ)K ′ if I $ K ′ . In order to show that
X(µ, b) → X(µ, b)K ′ is surjective, one needs to have some decomposition of
K ′ Adm(µ)K ′ , finer than the decomposition into K ′ double cosets. The idea of
the sought-after decomposition is essentially due to Lusztig. In [51], Lusztig in-
troduced G-stable pieces for reductive groups over algebraically closed fields. The
closure relation between G-stable pieces was determined in [23] and a more sys-
tematic approach using the “ partial conjugation action” technique was given later
in [24]. The notion and the closure relation of G-stable pieces also found appli-
cation in arithmetic geometry, e.g. in the work of Pink, Wedhorn and Ziegler on
algebraic zip data [58].

7.3. Ekedahl-Kottwitz-Oort-Rapoport stratification. Lusztig’s idea was also

adapted to loop groups, first with a hyperspecial parahoric subgroup, indepen-
dently by the author [25], and by Viehmann [70]. It was used it to define the
Ekedahl-Oort stratification of a general Shimura variety.
The desired decomposition of K ′ Adm(µ)K ′ for an arbitrary parahoric subgroup
K ′ was given in [27] as
K ′ Adm(µ)K ′ = ⊔w∈K′ W̃ ∩Adm(µ) K ′ ·σ IwI,
12 X. HE
′
where K W̃ is the set of minimal length elements in WK ′ \W̃ and ·σ means the σ-
conjugation action. This decomposition is used in joint work with Rapoport [32]
to define the Ekedahl-Kottwitz-Oort-Rapoport stratification of Shimura varieties
with arbitrary parahoric level structure. This stratification interpolates between
the Kottwitz-Rapoport stratification in the case of the Iwahori level structure and
the Ekedahl-Oort stratification [70] in the case of hyperspecial level structure.

8. Affine Deligne-Lusztig varieties with simple geometric

structure
8.1. Simple geometric structure for some X(µ, b0 )K ′ . The geometric struc-
ture of X(µ, b0 )K ′ for basic b0 is rather complicated in general. However, in certain
cases, X(µ, b0 )K ′ admit a simple description. The first nontrivial example is due to
Vollaard and Wedhorn in [73]. They showed that Xµ (b0 ) for an unramified unitary
group of signature (1, n − 1) and µ = (1, 0, · · · , 0) (and for hyperspecial parahoric
level structure), is a union of classical Deligne-Lusztig varieties, and the index set
and the closure relations between the strata are encoded in a Bruhat-Tits build-
ing. Since then, this question has attracted significant attention. We mention
the work of Rapoport, Terstiege and Wilson [62] on ramified unitary groups, of
Howard and Pappas [34], [35] on orthogonal groups, of Tiao and Xiao [66] in the
Hilbert-Blumenthal case. In all these works, the parahoric subgroups involved are
hyperspecial parahoric subgroups or certain maximal parahoric subgroups. The
analogous group-theoretic question for maximal parahoric subgroups was studied
in joint work with Görtz [15].
Note that these simple descriptions of closed affine Deligne-Lusztig varieties
(and the corresponding basic locus of Shimura varieties) have been used, with
great success, towards applications in number theory: to compute intersection
numbers of special cycles, as in the Kudla-Rapoport program [47] or in work [63],
[49] towards Zhang’s Arithmetic Fundamental Lemma [76]; and to prove the Tate
conjecture for certain Shimura varieties [67], [9].
The work of [73], [62], [34], [35], [66] focused on specific Shimura varieties with
certain maximal parahoric level structure. The work [15] studied the analogous
group-theoretic question for arbitrary reductive groups. The conceptual interpre-
tation on the occurrence of classical Deligne-Lusztig varieties was given; however,
a large part of the work in [15] was still obtained by brute force.
8.2. Some equivalent conditions. From the Lie-theoretic point of view, one
would like to consider not only the maximal parahoric subgroups, but all parahoric
subgroups; and one would like to have a conceptual understanding on the following
question:
When and why is X(µ, b0 )K ′ naturally a union of classical Deligne-Lusztig va-
rieties?
This was finally achieved in joint work with Görtz and Nie [16] as follows
Theorem 8.1. Assume that G is simple, µ is a dominant coweight of G and K ′
is a parahoric subgroup. Then the following conditions are equivalent:
• For basic [b0 ] ∈ B(G, µ), X(µ, b0 )K ′ is naturally a union of classical Deligne-
Lusztig varieties;
• For any nonbasic [b] ∈ B(G, µ), dim X(µ, b)K ′ = 0;
 13

• The pair (µ, b) is Hodge-Newton decomposable for any nonbasic [b] ∈ B(G, µ);
• The coweight µ is minute for G.
Here the minute condition is an explicit combinatorial condition on the coweight
P groups, it means that for any σ-orbit O on the set of simple
µ. For quasi-split
roots, we have i∈O hµ, ωi i 6 1. For non quasi-split groups, the condition is more
involved and we refer to [17, Definition 2.2] for the precise definition. It is also
worth mentioning that it is not very difficult to classify the pairs (G, µ) with the
minute condition. In [17, Theorem 2.5], a complete list of the cases is obtained,
where X(µ, b0 )K ′ is naturally a union of classical Deligne-Lusztig varieties.
Fargues and Rapoport conjectured that for p-adic period domains, the weakly
admissible locus coincides with the admissible locus if and only if the pair (µ, b) is
Hodge-Newton decomposable for any nonbasic [b] ∈ B(G, µ) (cf. [17, Conjecture
0.1]). This conjectured is established in a very recent preprint [4] by Chen, Fargues
and Shen.
8.3. Further remarks. From the Lie-theoretic point of view, there are some
quite striking new features in Theorem 8.1:
(1) The relations between the variety X(µ, b)K ′ for the basic σ-conjugacy class
and for nonbasic σ-conjugacy classes;
(2) The relation between the condition that X(µ, b0 )K ′ has a simple description
and the Hodge-Newton decomposability condition;
(3) The existence of a simple description of X(µ, b0 )K ′ is independent of the para-
horic subgroup K ′ .
Note that part (1) and part (2) are new even for the specific Shimura varieties
with hyperspecial level structure considered in the previous works. Part (3) is the
most mysterious one. In [17], we state that “We do not see any reason why this
independence of the parahoric could be expected a priori, but it is an interesting
parallel with the question when the weakly admissible and admissible loci in the
rigid analytic period domain coincide.”
For applications to number theory, one needs to consider the fraction field of the
Witt ring instead of the formal Laurent series field Fq ((ǫ)). In that setting, we have
a similar, but weaker result, namely, X(µ, b0 )K ′ is naturally a union of classical
Deligne-Lusztig varieties as perfect schemes. It is expected that the structural
results hold without perfection, as indicated in the special cases established in the
papers mentioned in §8.1.

9. Some applications to Shimura varieties

In the last subsection, we give a very brief discussion of some applications to
arithmetic geometry.
9.1. Some characteristic subsets. The study of some characteristic subsets in
the special fiber of a Shimura variety is a central topic in arithmetic geometry.
We mention the Newton strata, the Ekedahl-Oort strata for the hyperspecial level
structure and the Kottwitz-Rapoport strata for the Iwahori level structure. Con-
cerning these stratifications, there are many interesting questions one may ask,
e.g. which strata are nonempty, what is the relation between these various strati-
fications, etc.. These questions have been intensively studied in recent years and
14 X. HE

there is a large body of literature on these questions. Among them, we mention

the work of Viehmann and Wedhorn [72] on the nonemptiness of Newton strata
and Ekedahl-Oort strata for PEL type Shimura varieties with hyperspecial level
structure, the work of Kisin, Madapusi and Shin [39] on the nonemptiness of the
basic Newton stratum, the work of Hamacher [22] on the closure relation between
Newton strata, and the work of Wedhorn [74] and Moonen [55] on the density of
the µ-ordinary locus (i.e. the Newton stratum corresponding to ǫµ ). We refer to
[32, Introduction] and [71] for more references.
9.2. An axiomatic approach. In the works mentioned above, both algebro-
geometric and Lie-theoretic methods are involved, and are often mixed together.
In joint work with Rapoport [32], we purposed an axiomatic approach to the
study of these characteristic subsets in a general Shimura variety. We formulated
five axioms, based on the existence of integral models of Shimura varieties (which
have been established in various cases by the work of Rapoport and Zink [64],
Kisin and Pappas [40]), the existence of the following commutative diagram and
some compatibility conditions:
λK ′ 11 K ′ \G/K ′ .
rr rr99
rrrr
ℓ
rr K
ΥK ′
ShK ′ // G/K ′ σ
▲▲▲
▲▲d▲K ′
▲▲▲
▲&&
δK .. B(G)

Here K ′ is a parahoric subgroup, ShK ′ is the special fiber of a Shimura variety

with K ′ level structure, and G/Kσ′ is the set-theoretic quotient of G by the σ-
conjugation action of K ′ .
As explained in [17, §6.2], affine Deligne-Lusztig varieties are involved in the
diagram in an essential way, via the bijection
∼
Jb \X(µ, b)K ′ −→ d−1 −1 ′ ′
K ′ ([b]) ∩ ℓK ′ (K Adm({µ})K ).

9.3. Some applications and current status of the axioms. It is shown in

[32] that under those axioms, the Newton strata, the Ekedahl-Oort strata, the
Kottwitz-Rapoport strata, and the Ekedahl-Kottwitz-Oort-Rapoport strata dis-
cussed in §7, are all nonempty in their natural range. Furthermore, under those
axioms several relations between these various stratifications are also established
in [32].
Following [32], Shen and Zhang in [65] studied the geometry of good reductions
of Shimura varieties of abelian type. They established basic properties of these
characteristic subsets, including nonemptiness, closure relations and dimension
formula and some relations between these stratifications.
In joint work with Nie [31], based on the framework of [32], we studied the
density problem of the µ-ordinary locus. Under the axioms of [32] we gave several
explicit criteria on the density of the µ-ordinary locus.
Algebraic geometry is essential in the verification of these axioms. For PEL
type Shimura varieties associated to unramified groups of type A and C and to
odd ramified unitary groups, the axioms are verified in joint work with Zhou [33].
 15

For Shimura varieties of Hodge type, most of the axioms are verified in recent
work of Zhou [77].

References
[1] B. Bhatt and P. Scholze, Projectivity of the Witt vector Grassmannian, Invent. math. 209
(2017), 329–423.
[2] F. Bruhat and J. Tits, Groupes rèductifs sur un corps local, I, Publ. Math. IHES 41 (1972),
5–276.
[3] C.-L. Chai, Newton polygons as lattice points, Amer. J. Math. 122 (5) (2000), 967–990.
[4] M. Chen, L. Fargues and X. Shen, On the structure of some p-adic period domains,
arXiv:1710.06935.
[5] M. Chen, M. Kisin and E. Viehmann, Connected components of affine Deligne-Lusztig va-
rieties in mixed characteristic, Compositio Math. 151 (2015), 1697–1762.
[6] L. Chen and S. Nie, Connected components of closed affine Deligne-Lusztig varieties,
arXiv:1703.02476.
[7] D. Ciubotaru and X. He, Cocenters and representations of affine Hecke algebra, J. Eur.
Math. Soc. 19 (2017), 3143–3177.
[8] P. Deligne and G. Lusztig, Representations of reductive groups over finite fields, Ann. of
Math. (2) 103 (1976), 103–161.
[9] D. Helm, Y. Tian, L. Xiao, Tate cycles on some unitary Shimura varieties mod p, Algebra
Number Theory 11 (2017), 2213–2288.
[10] D. Gaitsgory, Construction of central elements in the affine Hecke algebra via nearby cycles,
Invent. Math. 144 (2001), 253–280.
[11] Q. Gashi, On a conjecture of Kottwitz and Rapoport, Ann. Sci. École Norm. Sup. (4) 43
(2010), no. 6, 1017–1038.
[12] M. Geck and G. Pfeiffer, On the irreducible characters of Hecke algebras, Adv. Math. 102
(1993), 79–94.
[13] U. Görtz, T. Haines, R. Kottwitz, D. Reuman, Dimensions of some affine Deligne-Lusztig
varieties, Ann. sci. École Norm. Sup. 39 (2006), 467–511.
[14] U. Görtz, T. Haines, R. Kottwitz, D. Reuman, Affine Deligne-Lusztig varieties in affine
flag varieties, Compos. Math.146 (2010), 1339-1382.
[15] U. Görtz and X. He, Basic loci in Shimura varieties of Coxeter type, Cambridge J. Math. 3
(2015), no. 3, 323–353.
[16] U. Görtz, X. He and S. Nie, P -alcoves and nonemptiness of affine Deligne-Lusztig varieties,
Ann. Sci. Ècole Norm. Sup. 48 (2015), 647–665.
[17] U. Görtz, X. He and S. Nie, Fully Hodge-Newton decomposable Shimura varieties, arXiv:
1610.05381.
[18] U. Görtz and C.-F. Yu, Supersingular Kottwitz-Rapoport strata and Deligne-Lusztig vari-
eties, J. Inst. Math. Jussieu 9 (2010), 357–390.
[19] T. Haines and X. He, Vertexwise criteria for admissibility of alcoves, Amer. J. Math. 139
(2017), 769–784.
[20] T. Haines, B. C. Ngô, Alcoves associated to special fibers of local models, Amer. J. Math.
124 (2002), 1125-1152.
[21] P. Hamacher, The dimension of affine Deligne-Lusztig varieties in the affine Grassmannian
of unramified groups, Int. Math. Res. Notices 23 (2015), 12804–12839.
[22] P. Hamacher, The geometry of Newton strata in the reduction modulo p of Shimura varieties
of PEL type, Duke Math. J. 164 (2015), 2809–2895.
[23] X. He, The G-stable pieces of the wonderful compactification, Trans. Amer. Math. Soc., 359
(2007), 3005–3024.
[24] X. He, Minimal length elements in some double cosets of Coxeter groups, Adv. Math. 215
(2007), no. 2, 469–503.
[25] X. He, Closure of Steinberg fibers and affine Deligne-Lusztig varieties, Int. Math. Res. No-
tices 14 (2011), 3237–3260.
16 X. HE

[26] X. He, Geometric and homological properties of affine Deligne-Lusztig varieties, Ann. Math.
179 (2014), 367–404.
[27] X. He, Kottwitz-Rapoport conjecture on unions of affine Deligne-Lusztig varieties, Ann. Sci.
Ècole Norm. Sup. 49 (2016), 1125–1141.
[28] X. He, Hecke algebras and p-adic groups, Current developments in mathematics 2015, 73–
135, Int. Press, Somerville, MA, 2016.
[29] X. He and S. Nie, Minimal length elements of extended affine Weyl group, Compos. Math.
150 (2014), no. 11, 1903–1927.
[30] X. He and S. Nie, On the acceptable elements, IMRN 2018 (2018), 907–931.
[31] X. He and S. Nie, On the µ-ordinary locus of a Shimura variety, Adv. Math. 321 (2017),
513–528.
[32] X. He and M. Rapoport, Stratifications in the reduction of Shimura varieties, Manuscripta
Math. 152 (2017), 317–343.
[33] X. He and R. Zhou, On the connected components of affine Deligne-Lusztig varieties,
arXiv:1610.06879.
[34] B. Howard and G. Pappas, On the supersingular locus of the GU (2, 2) Shimura variety,
Algebra Number Theory 8 (2014), no. 7, 1659–1699.
[35] B. Howard, G. Pappas, Rapoport-Zink spaces for spinor groups, Compos. Math. 153 (2017),
1050–1118.
[36] N. Iwahori and H. Matsumoto, On some Bruhat decomposition and the structure of the
Hecke rings of p-adic Chevalley groups, Publ. Math. IHES 25 (1965), 5–48.
[37] H. Katz, Slope filtration of F -crystals, Astérisque 63 (1979), 113–163.
[38] M. Kisin, Mod-p points on Shimura varieties of abelian type, J. Amer. Math. Soc. 30 (2017),
819–914.
[39] M. Kisin, K. Madapusi, S.-W. Shin, Honda-Tate theory for Shimura varieties of Hodge type,
in preparation.
[40] M. Kisin and G. Pappas, Integral models of Shimura varieties with parahoric level structure,
arXiv:1512.01149, to appear in Publ. Math. IHES.
[41] R. Kottwitz, Isocrystals with additional structure, Compositio Math. 56 (1985), 201–220.
[42] R. Kottwitz, Isocrystals with additional structure. II, Compositio Math. 109 (1997), 255–
339.
[43] R. Kottwitz, On the Hodge-Newton decomposition for split groups, IMRN 26 (2003), 1433–
1447.
[44] R. Kottwitz, Dimensions of Newton strata in the adjoint quotient of reductive groups, Pure
Appl. Math. Q. 2 (2006), 817–836.
[45] R. Kottwitz, M. Rapoport, Minuscule alcoves for GLn and GSP2n , Manuscripta Math.
102, no.4, (2000), 403-428.
[46] R. Kottwitz and M. Rapoport, On the existence of F -crystals, Comment. Math. Helv., 78
(2003), 153–184.
[47] S. Kudla and M. Rapoport, Special cycles on unitary Shimura varieties I. Unramified local
theory, Invent. Math. 184 (3) (2011), 629–682.
[48] R. Langlands, M. Rapoport, Shimuravarietäten und Gerben, J. Reine Angew. Math 378
(1987), 113–220.
[49] C. Li and Y. Zhu, Remarks on the arithmetic fundamental lemma, Algebra Number Theory
11 (2017), 2425–2445.
[50] G. Lusztig, Characters of reductive groups over a finite field, Annals of Mathematics Studies,
107. Princeton University Press, Princeton, NJ, 1984.
[51] G. Lusztig, Parabolic character sheaves, I, II, Mosc. Math. J. 4 (2004), no. 1, 153–179; no. 4,
869–896.
[52] G. Lusztig, From conjugacy classes in the Weyl group to unipotent classes, I–III, Repre-
sent.Th. 15(2011), 494-530; 16(2012), 189–211, 450-488.
[53] G. Lusztig, Bruhat decomposition and applications, arXiv:1006.5004.
[54] B. Mazur, Frobenius and the Hodge filtration (estimates), Ann. of Math. (2) 98 (1973),
58–95.
 17

[55] B. Moonen, Serre-Tate theory for moduli spaces of PEL-type, Ann. Sci. Ècole Norm. Sup.
37 (2004), 223–269.
[56] S. Nie, Connected components of closed affine Deligne-Lusztig varieties in affine Grassman-
nians, arXiv:1511.04677.
[57] G. Pappas and M. Rapoport, Twisted loop groups and their affine flag varieties, Adv. Math.
219 (2008), 118–198, with an appendix by T. Haines and Rapoport.
[58] R. Pink, T. Wedhorn, and P. Ziegler, Algebraic zip data Doc. Math. 16 (2011), 253–300.
[59] E. A. Rad and U. Hartl, Langlands-Rapoport Conjecture Over Function Fields,
arXiv:1605.01575.
[60] M. Rapoport, A guide to the reduction modulo p of Shimura varieties, Astérisque (2005),
no. 298, 271–318.
[61] M. Rapoport and M. Richartz, On the classification and specialization of F -isocrystals with
additional structure, Compositio Math. 103 (1996), no. 2, 153–181.
[62] M. Rapoport, U. Terstiege, S. Wilson, The supersingular locus of the Shimura variety for
GU (1, n − 1) over a ramified prime, Math. Z. 276 (2014), 1165–1188.
[63] M. Rapoport, U. Terstiege, W. Zhang, On the Arithmetic Fundamental Lemma in the
minuscule case, Compos. Math. 149 (2013), 1631–1666.
[64] M. Rapoport, Th. Zink, Period spaces for p-divisible groups, Annals of Mathematics Studies,
141. Princeton University Press, Princeton, NJ, 1996.
[65] X. Shen and C. Zhang, Stratifications in good reductions of Shimura varieties of abelian
type, arXiv:1707.00439.
[66] Y. Tian, L. Xiao, On Goren-Oort stratification for quaternionic Shimura varieties, Compos.
Math. 152 (2016), 2134–2220.
[67] Y. Tian, L. Xiao, Tate cycles on quaternionic Shimura varieties over finite fields,
arXiv:1410.2321 (2014).
[68] E. Viehmann. The dimension of affine Deligne-Lusztig varieties, Ann. Sci. École Norm.
Sup. (4) 39 (2006), 513–526.
[69] E. Viehmann, Connected components of affine Deligne-Lusztig varieties, Math. Ann. 340
(2008), 315–333.
[70] E. Viehmann, Truncations of level 1 of elements in the loop group of a reductive group, Ann.
of Math. (2) 179 (2014), 1009–1040.
[71] E. Viehmann, On the geometry of the Newton stratification, arXiv:1511.03156.
[72] E. Viehmann, T. Wedhorn, Ekedahl-Oort and Newton strata for Shimura varieties of PEL
type, Math. Ann. 356 (2013), 1493–1550.
[73] I. Vollaard, T. Wedhorn, The supersingular locus of the Shimura variety of GU(1,n-1) II,
Invent. Math. 184 (2011), 591–627.
[74] T. Wedhorn, Ordinariness in good reductions of Shimura varieties of PEL-type, Ann. Sci.
Ècole Norm. Sup. (4), 32 (1999), 575–618.
[75] J.-P. Wintenberger, Existence de F -cristaux avec structures supplémentaires, Adv. Math.
190 (2005), 196–224.
[76] W. Zhang, On arithmetic fundamental lemmas, Invent. Math., 188 (2012), 197–252.
[77] R. Zhou, Mod-p isogeny classes on Shimura varieties with parahoric level structure, arXiv:
1707.09685.
[78] X. Zhu, On the coherence conjecture of Pappas and Rapoport, Ann. of Math. (2) 180 (2014),
1–85.
[79] X. Zhu, Affine Grassmannians and the geometric Satake in mixed characteristic, Ann. of
Math. (2) 185 (2017), 403–492.

Department of Mathematics, University of Maryland, College Park, MD 20742

E-mail address: [email protected]