Workshop on the arithmetic geometry of Shimura varieties,

representation theory, and related topics

[Link]

9:30–10:30
Rapoport-Zink spaces and endoscopy
10:45–11:45
Geometric realization of the local Langlands correspondence for
representations of conductor three
13:45–14:45 Keerthi Madapusi Pera (Harvard)
(Talk 1) Integral canonical models for Shimura varieties of Hodge type
15:15–16:15 Gabriel Dospinescu (Ecole Polytechnique)
(Talk 1) Overview of the p-adic Langlands correspondence for GL2 (Qp )
16:30–17:30
A generalization of Kato’s local ε conjecture for (φ, Γ)-modules over
the Robba ring

9:30–10:30 Florian Herzig (Toronto)

(Talk 1) Ordinary representations of GLn (Qp ) and fundamental
algebraic representations, I
10:45–11:45 Keerthi Madapusi Pera (Harvard)
(Talk 2) Regular models for Shimura varieties of orthogonal type
and the Tate conjecture for K3 surfaces in odd characteristic
13:45–14:45 Gabriel Dospinescu (Ecole Polytechnique)
(Talk 2) Locally analytic Jacquet modules and p-adic differential
equations
15:15–16:15 Kai-Wen Lan (Minnesota)
(Talk 1) Compactifications of PEL-type Shimura varieties and Kuga
families with ordinary loci
16:30–17:30
Good reduction criterion for K3 surfaces
9:30–10:30 Florian Herzig (Toronto)
(Talk 2) Ordinary representations of GLn (Qp ) and fundamental
algebraic representations, II
10:45–11:45 Benjamin Schraen (CNRS/UVSQ)
(Talk 1) Completed cohomology of the tower of modular curves
and local-global compatibility
12:00–13:00 Tasho Kaletha (Princeton/IAS)
(Talk 1) Construction of supercuspidal L-packets for tamely-ramified
groups, I
Discussion/Free Afternoon

9:30–10:30 Benjamin Schraen (CNRS/UVSQ)

(Talk 2) The cohomology of local systems on quotient of Drinfel’d
upper half spaces
10:45–11:45 Tasho Kaletha (Princeton/IAS)
(Talk 2) Construction of supercuspidal L-packets for tamely-ramified
groups, II
13:45–14:45 Kai-Wen Lan (Minnesota)
(Talk 2) Galois representations for regular algebraic automorphic forms
over CM fields
15:15–16:15 Przemyslaw Chojecki (Université Pierre et Marie Curie)
Mod p non-abelian Lubin-Tate theory
16:30–17:30 Xu Shen (Orsay)
Cell decomposition and Lefschetz trace formula for some unitary
group Rapoport-Zink spaces

9:30–10:30
Potentially good reduction loci of open Shimura varieties and their
ℓ-adic cohomology
10:45–11:45
Conjecture on the stable reduction of the Lubin-Tate curve and
the Lusztig theory over finite rings
12:00–13:00 Sug Woo Shin (MIT)
Integral Manin problem with PEL structure
 Workshop on the arithmetic geometry of Shimura varieties,
representation theory, and related topics

Przemyslaw Chojecki (Université Pierre et Marie Curie)

Mod p non-abelian Lubin-Tate theory
I will show that the mod p etale cohomology of the Lubin-Tate tower realizes
the mod p local Langlands correspondence and a mod p Jacquet-Langlands correspondence. This
will be done by comparing the cohomology of the Lubin-Tate tower with the cohomology of the
tower of modular curves and using results of Emerton with some further representation-theoretic
analysis. I will show two possible notions of a mod p Jacquet-Langlands correspondence and I will
discuss a connection of them with a local-global compatibility part of the Buzzard-Diamond-Jarvis
conjecture.

Gabriel Dospinescu (Ecole Polytechnique)

(Talk 1) Overview of the p-adic Langlands correspondence for GL2 (Qp )
We will try to recall in an accessible manner the main objects and results in the
p-adic Langlands correspondence, due to Colmez. We will also try to relate them to more recent
work of Paskunas.
(Talk 2) Locally analytic Jacquet modules and p-adic differential equations
We will show that simple p-adic differential equations control the Jacquet mod-
ules of the unitary p-adic principal series of GL2 (Qp ). This leads to a very direct proof of conjec-
tures of Berger, Breuil and Emerton. We will try to discuss the relationship between our approach
and that of Liu, Xie, Zhang and Colmez.

Florian Herzig (Toronto)

(Talk 1 & 2) Ordinary representations of GLn (Qp ) and fundamental algebraic
representations
Motivated by a hypothetical p-adic Langlands correspondence for GLn (Qp ) we
associate to an n-dimensional ordinary (i.e. upper-triangular) representation ρ of Gal(Qp /Qp ) over
E a unitary Banach space representation Π(ρ)ord of GLn (Qp ) over E that is built out of principal
series representations. (Here, E is a finite extension of Qp .) There is an analogous construction
over Fp . In the latter case we show under suitable hypotheses that Π(ρ)ord occurs in the ρ-part
of the cohomology of a compact unitary group. This is joint work with Christophe Breuil. In the
first talk we will discuss some relevant background on mod p and p-adic Langlands, Serre weights,
and mod p smooth representations of p-adic groups.
1
 Geometric realization of the local Langlands correspondence for represen-
tations of conductor three
We will prove that the local Langlands correspondence for two-dimensional rep-
resentations of Weil groups of conductor three is realized in the cohomology of one-dimensional
Lubin-Tate spaces by studying stable models of Lubin-Tate spaces. In this talk, we will put
emphasis on the case where a base field is dyadic. This is a joint work with Takahiro Tsushima.

Rapoport-Zink spaces and endoscopy

Rapoport-Zink spaces are generalizations of the Lubin-Tate spaces and the Drin-
feld upper half spaces, whose ℓ-adic cohomology is expected to realize the local Langlands corre-
spondences and local Jacquet-Langlands correspondences. It is proved by Harris-Taylor, Boyer,
Faltings, Fargues,... for the Lubin-Tate spaces and the Drinfeld upper half spaces, but the general
case is widely open. Accoring to the work of Boyer for GL(n), and the speaker and Yoichi Mieda
for GSp(4) and GU(3), it turned out that a naive expectation should be modified in order to treat
local A-packets and L-packets. We try to discuss possible description of the endoscopic decomposi-
tion of the ℓ-adic cohomology of the Rapoport-Zink spaces (“Mirror Symmetry for Rapoport-Zink
spaces”). This talk also aims to introduce basic notations and definitions on Shimura varieties,
Rapoport-Zink spaces, and local/global Langlands correspondences which appear in other lectures
in this workshop.

Tasho Kaletha (Princeton/IAS)

(Talk 1 & 2) Construction of supercuspidal L-packets for tamely-ramified
groups
We will report on recent work to construct L-packets on general tamely-ramified
reductive p-adic groups consisting of supercuspidal representations of low ramification. In the first
talk, we will recall some background material on endoscopy and state precise versions of the local
Langlands conjecture and endoscopic transfer conjecture, which are adapted to groups that are
not necessarily quasi-split. In the second talk, we will state what is currently known for general
groups and discuss some of the details of the construction.

Kai-Wen Lan (Minnesota)

(Talk 1) Compactifications of PEL-type Shimura varieties and Kuga fam-
ilies with ordinary loci
I will explain the constructions of normal flat p-integral models of various al-
gebraic compactifications of PEL-type Shimura varieties and Kuga families, allowing ramification
(including levels) at p, with good behaviors over the loci where certain (multiplicative) ordinary
level structures are defined. (Explicitly, in the case of GSp(2n)/totally real or GU(n, n)/CM, for

2
example, we allow arbitrary ramifications of p in the totally real or CM field, and allow level
structures as deep as Γ1 (pm ) for arbitrary m.) These constructions (in the case of GU(n, n)/CM)
are used in my joint work with Michael Harris, Richard Taylor, and Jack Thorne on attaching
overconvergent cusp forms and Galois representations to regular algebraic automorphic represen-
tations of GL(n) over CM fields, without any hypotheses on self-duality.
(Talk 2) Galois representations for regular algebraic automorphic forms
over CM fields
I will report on my joint work with Michael Harris, Richard Taylor, and Jack
Thorne on the construction of p-adic Galois reprensentations for regular algebraic cuspidal au-
tomorphic representations of GL(n) over CM (or totally real) fields, without any self-duality
hypothesis.
I will focus on a crucial new idea, namely the construction of overconvergent cusp forms with
desired Hecke actions using the partial toroidal boundary of the ordinary loci of unitary Shimura
varieties and Kuga families over them. I will outline other needed steps for the construction of the
desired Galois representations (for automorphic representations of GL(n)) using such overconver-
gent cusp forms, and give more details if time permits.
I would like to emphasize that it is believed that such Galois reprensentations do not appear in
the p-adic etale cohomology of Shimura varieties and Kuga families over them, at least not in a
way detectable by counting points in characteristic p. Instead, we construct such Galois repre-
sentations using p-adic limits of known ones, and this is why the construction of overconverngent
cusp forms with desired Hecke actions is crucial, as they can be viewed as p-adic limits of classical
cusp forms (for which people do know how to attach Galois representations, thanks to previous
works with self-duality hypotheses).

Keerthi Madapusi Pera (Harvard)

(Talk 1) Integral canonical models for Shimura varieties of Hodge type
I will give an overview of the construction of integral models for Shimura vari-
eties of Hodge type via the methods of Kisin and Faltings (as well as Vasiu). Since these Shimura
varieties have no moduli interpretation, the methods are necessarily ad hoc, involving Deligne’s
theory of absolute Hodge cycles, p-adic Hodge theory and the deformation theory for p-divisible
groups. I will quickly review these, and indicate the key difficulties in the construction.
(Talk 2) Regular models for Shimura varieties of orthogonal type and the
Tate conjecture for K3 surfaces in odd characteristic
I will show how to construct good integral models for orthogonal Shimura va-
rieties even at (odd) primes where the discriminant is not invertible. Then we will see that the
certain moduli spaces of polarized K3 surfaces can be viewed as open sub-schemes of these integral
models. This, combined with recent results of Kisin towards the Langlands-Rapoport conjecture,
will lead us to a proof of the Tate conjecture in odd characteristic p for K3 surfaces admitting
a polarization of degree indivisible by p2 . The same methods also prove the Tate conjecture for
cubic fourfolds in odd characteristic.

3
 Good reduction criterion for K3 surfaces
Abelian varieties over local fields have good reduction if and only if their l-adic
etale cohomology (equivalently Tate modules) are unramified [Serre–Tate]. I prove a similar result
for (potential) good reduction of projective K3 surfaces of small degree (relative to the residue
characteristic). This result covers wider classes of K3 surfaces than those in my previous talks (in
Kyoto, Hiroshima, Kyushu).

Potentially good reduction loci of open Shimura varieties and their ℓ-adic
cohomology
If a Shimura variety is compact and has a proper integral model, its ℓ-adic
cohomology can be computed by the nearby cycle complex. However, if it is not compact, we
do not have such a result in general. In this talk, for a certain class of Shimura varieties, I will
introduce rigid open subsets called the potentially good reduction loci. If a Shimura variety has a
reasonable integral model, then the ℓ-adic cohomology of this open subset will coincide with the
cohomology of the nearby cycle complex. We compare the compactly supported ℓ-adic cohomology
of a Shimura variety and that of its potentially good reduction locus. This is a joint work with
Naoki Imai.

A generalization of Kato’s local ε conjecture for (φ, Γ)-modules over the

Robba ring
In his preprint “Lectures on the approach to Iwasawa theory of Hasse-Weil L-
functions via BdR , Part II”, Kazuya Kato proposed a conjecture which is called local ε conjecture.
This conjecture roughly says that the determinant of the Galois cohomology of a family of p-adic
Galois representations has a canonical base which can be described by using Deligne-Langlands
ε constants. In our talk, we generalize his conjecture for families of (φ, Γ)-modules over the
Robba ring and prove a part of this conjecture in some special cases (more precisely, trianguline
case). The two key ingredients are the recent result of Kedlaya-Pottharst-Xiao on the finiteness of
cohomologies of (φ, Γ)-modules and my result on Bloch-Kato exponential maps for (φ, Γ)-modules.
This talk is a work in progress.

Benjamin Schraen (CNRS/UVSQ)

(Talk 1) Completed cohomology of the tower of modular curves and local-
global compatibility
We will introduce completed p-adic etale cohomology of the tower of modular
curves. We will next explain the strategy of Emerton to realize the local p-adic Langlands corre-
spondence for GL2(Qp) in this cohomology.
(Talk 2) The cohomology of local systems on quotient of Drinfel’d upper
half spaces
4
 (joint work with Jan Kohlhaase) If X is a d-dimensional variety uniformized by
Drinfel’d upper half space, Schneider has conjectured the transversality of two natural filtrations
on the de Rham cohomology of local systems on X. We will explain how vanishing theorems for
the cohomology of discrete groups on locally analytic representations of GLd+1 could be used to
prove this conjecture, and effectively prove any new cases.

Xu Shen (Orsay)
Cell decomposition and Lefschetz trace formula for some unitary group
Rapoport-Zink spaces
We explain that, how to use Fargues’s theory of Harder-Narasimhan filtration
for finite flat group schemes to find certain compact analytic domain in the basic Rapoport-Zink
spaces of signature (1, n − 1) for some unitary groups, so that its translations under the two
related p-adic reductive groups form a locally finite covering of the whole space. By studying the
action of regular elliptic elements and applying Mieda’s theorem, we can deduce a Lefschetz trace
formula for all large enough analytic domains coming from this covering. This will be sufficient
for applications to representation theory.

Sug Woo Shin (MIT)

Integral Manin problem with PEL structure
The rational/integral Manin problem asks whether a given p-divisible group
over the algebraic closure of Fp can be realized as the p-divisible group of an abelian variety over
the same field up to an isogeny/isomorphism. The analogous problem for p-divisible groups and
abelian varieties with PEL structure have been solved by Vasiu and Viehmann-Wedhorn for p-
unramified PEL data of type A and C. I will explain a different approach based on Honda-Tate
theory and Galois cohomology to obtain the answer without any unramified hypothesis. As an
application one obtains a relatively simple argument to show the non-emptiness of Newton strata
of Shimura varieties attached to p-unramified PEL data of type A and C.

Conjecture on the stable reduction of the Lubin-Tate curve and the Lusztig
theory over finite rings
A connection between the non-abelian Lubin-Tate theory and the Deligne-
Lusztig theory seems to be mysterious. T. Yoshida proves that the Deligne-Lusztig variety for
GLh appears in the stable reduction of the Lubin-Tate space of level 1.
In this talk, we construct a curve over a finite field with some group action using Lusztig theory
over finite rings and study its cohomology. We state a conjecture that this curve appears in the
stable reduction of the Lubin-Tate curve. As an evidence, we prove this conjecture in the case
where the level is two. This is a joint work with Tetsushi Ito and Yoichi Mieda.