• 検索結果がありません。

Conditional Cyclic Base Change (Automorphic Forms and Number Theory)

N/A
N/A
Protected

Academic year: 2021

シェア "Conditional Cyclic Base Change (Automorphic Forms and Number Theory)"

Copied!
6
0
0

読み込み中.... (全文を見る)

全文

(1)

Conditional

Cyclic

Base Change

Jean-Pierre LABESSE

1 – Introduction

Base change is a major topic in the modern theory of automorphic form.

The quadratic base change for modular forms, from $\mathbb{Q}$ to a real quadratic

field $\mathbb{Q}(\sqrt{d})$, was first obtained by Doi and Naganuma using L-functions

[DN]. Then Saito [Sa] discovered that cyclic base change could be

de-duced from the comparison between a trace formula and a twisted trace

formula; but technical limitations to his result arose from the fact he was

working in the classical language. Shintani reformulated Saito’s method

in the representations theoretic and adelic framework and understood

it should yield unconditional results for the cyclic base change of

auto-morphic representations for $GL(2)$; this was worked out by Langlands

[Lan]. From this and other quite deep results (due to Gelbart, Jacquet,

Piatetskii-Shapiro and Shalika), Langlands deduced non trivial cases of

Artin’s conjecture. This last result, as completed by Tunnell, is an

essen-tial step in Wiles’ proof of Fermat last theorem.

Some further important contributions should be quoted. In [AC],

Arthur and Clozel proved the existence of base change for $GL(n)$, and

Rogawski [Rog] established base change between unitary groups in 3

vari-ables and their split (outer) form $GL(3)$

.

Partial results have also been

obtained for more general unitary groups (see below).

In this note I would like to describe how much one can prove as

(2)

cannot get

an

unconditional theorem for arbitrary reductive groups since

a key ingredient in the proof-the so-called fundamental lemma-is only

available (but for a few exceptions) for the endoscopic group attached to

the trivial endoscopic character. To circumvent this lack of knowledge

we have to restrict to representations of a certain type at some specific

places.

2 –Weak base change

Let $E/F$ be a cyclic extension of a number field $F$, and let $\theta$ be a generator

of the Galois group $\mathrm{G}\mathrm{a}1(E/F)$.

Let $G_{0}$ be a reductive group over $F$; let $G$ be the restriction of scalars

from $E$ to $F$ of $G_{0}$; this is the group scheme such that for any F-algebra

$A$

$G(A)=G_{0}(E\otimes_{F}A)$

Observe that $\theta$ defines an automorphism of $G$ over $F$. Denote by $H$

the quasi-split inner form of $G_{0}$

.

The group $H$ is the endoscopic group

attached to the trivial endoscopic character for $G$ twisted by $\theta$.

For almost all places $v$ of $F$, the group scheme $G$ is defined over $\mathrm{O}_{v}$

the ring of integers of $F_{v}$; then we consider $K_{v}^{G}=G(D_{v})$

.

For almost all

places $v$ the groups $K_{v}^{G}$ and $K_{v}^{H}$ are

h.yperspecial

maximal compact

sub-groups in $G_{v}:=G(F_{v})$ and $H_{v}$ and one may choose Borel $F_{v}$-subgroups

$B_{v}^{G}\subset G_{v}$ and $B_{v}^{H}\subset H_{v}$.

For such places we have the notion of unramified representations

i.e. irreducible admissible representations with a non zero invariant

vec-tor under the hyperspecial maximal compact subgroup. Recall that an

unramified representations is the unique subquotient with a non zero

in-variant vector under the hyperspecial maximal compact subgroup, of a

principal series representations induced by an unramified characters of

the Borel subgroup.

There is a norm map between the tori quotients of the Borel

sub-groups $B_{v}^{G}$ and $B_{v}^{H}$ by their unipotent radicals. For tori, base change is

nothing but the composition of characters with the norm map. For

un-ramified representation the local base change is defined using local base

change for characters of $\mathrm{t}\mathrm{h}\mathrm{e}\backslash ...$

.

tori, $\mathrm{q}\mathrm{u}\mathrm{o}.\mathrm{t}\mathrm{i}\mathrm{e}\mathrm{n}\backslash \mathrm{t}\mathrm{s}$ of the $\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{e}1,$

$.\mathrm{S}\mathrm{u}\mathrm{b}\mathrm{g}\mathrm{r}\mathrm{o}\mathrm{u}_{\mathrm{P}^{\mathrm{S}}}$

. by

their unipotent radicals.

Definition. Let $\pi_{v}^{H}$ be an unramified representation$s$ of$H(F_{v})$ that is a

(3)

$\eta_{v}$ ofaBorel $F$-subgroup $B_{v}^{H}$. We say that $\pi_{v}^{G}$ is the local base chan$ge$ of

$\pi_{v}^{H}$ if$\pi_{v}^{G}$ is the subquotient with a

$\mathrm{n}$on zero invariant vector under $K_{v}^{G}$

of the principal series representation induced by the character

$\xi_{v}=\eta_{v^{\mathrm{O}}E/}NF$

We shall not try to define local base change in general. Unramified

local base change is enough to define weak (global) base change:

Definition. Let $\pi^{H}$ an$d\pi^{G}$ be $\mathrm{t}wo$ admissible irreducible

representa-tions of the adelic groups $H(\mathrm{A}_{F})$ and $G(\mathrm{A}_{F})$. We say that $\pi^{G}$ is $a$ weak

$b$as$\mathrm{e}$ change of

$\pi^{H}$ if for almost all places

$v$ the

rep.resentatio.

$n\pi_{v}^{c}$ is th$\mathrm{e}$

local base change of$\pi_{v}^{H}$

3 $-\theta$-stable representations

Let $\pi^{G}$ be an admissible irreducible representation of

$G(\mathrm{A}_{F})$ which is

$\theta$-stable

$\pi^{G_{\mathrm{o}\theta\simeq\pi}c}$

.

Denote by $A(\pi^{G})$ the isotypic component of$\pi^{G}$ in the cuspidal spectrum.

The automorphism $\theta$ acts on $A(\pi^{G})$ and we thus get a

natural

represen-tation of the semi-direct product $G(\mathrm{A}_{F})\mathrm{x}\mathrm{G}\mathrm{a}1(E/F)$ in this space.

Definition. We say that a cuspidal automorphic representation $\pi^{G}$

con-tributes non trivially to the $\theta$-twist$\mathrm{e}d$ trace formula if th

$\mathrm{e}$ trace of opera-$\mathrm{t}ors$ on $A(\pi^{G})$ defined by smooth compactly supported functions on the

coset $G(\mathrm{A}_{F})x\theta$, is not identically zero.

Observe that $\theta$-stable cuspidal representations

that occur with

mul-tiplicity one contributes non trivially to the $\theta$-twisted trace formula.

4 –The main theorems

It would be too technical to state the results for the most general

situ-ation; we shall assume that the derived group $H_{der}$ is $\mathrm{s}$

.imply

connected

and that the $\mathrm{c}\mathrm{o}$-center $D_{H}=H/H_{der}$ is a torus split over an extension

$E_{0}$ of $F$ contained in $E$. In particular $G$ satisfies the Hasse principle. Let

$\mathfrak{B}$ be a finite set of finite places such that for at least one place $v\in \mathfrak{B}$

the algebra $F_{v}\otimes E_{0}$ is a field. We assume moreover that $\mathfrak{B}$ contains at

(4)

Theorem 1. Let $\pi^{G}=\otimes\pi_{v}^{G}$ be an automorphic cuspidal representation

for$G$ over$F$ such that for all$v\in \mathfrak{B}$ the representation $\pi_{v}^{G}$ is the Steinberg

representation and assume that the representation $\pi^{G}$ contributes non

$\mathrm{t}$rivially to the $\theta$-twisted trace formula. Then, there exist an automorphic

cuspid$\mathrm{a}\iota_{repr}eSent..\mathrm{a}$tion $\pi^{H}$ for $H$ such that

$\pi^{G}$ is a

weak..b

$aSe=$ chan$ge$ of

$\pi^{H}$.

In the particular case where $E=F$ we obtain a (conditional) transfer

for cuspidal representations from an inner form to its quasi split form “\‘a

la Jacquet-Langlands” For the

converse

theorem, i.e. the lifting theorem

we have to assume that $G_{0}$ is quasi-split: $G_{0}=H$

.

Theorem 2. Assume that $G_{0}$ is $qu\mathrm{a}si$-spli$\mathrm{t}$. Let $\pi^{H}$ be an

automor-phic cuspidal representation for $G$ over $F$ such that for all $v\in \mathfrak{B}$ th$\mathrm{e}$

representation $\pi_{v}^{H}$ is the $S\mathrm{t}$einb

$\mathrm{e}r\mathrm{g}$ representation. Then, there exist an

autom$\mathit{0}$rphic cuspid

$\mathrm{a}l$ representation $\pi^{G}$ for $G$ such that $\pi^{G}$ is a weak

$b$as$\mathrm{e}$ chan$ge$ of

$\pi^{H}$.

5 –Application to cuspidal cohomology

The compatibility of weak base change with the local base change, when

defined, can be checked in many cases. This is in particular the case for

representation with cohomology: one can show that a cuspidal

represen-tation as in theorem 2, with non trivial cohomology at some places has

a weak base change with non trivial cohomology at those places.

Let $F$ be a number field and let $S$ be a finite set of places containing

all archimedean ones. Consider an $S$-arithmetic subgroup $\Gamma$ of

$G_{S}:=G(F_{S})$

.

One defines its cuspidal cohomology as follows:

$H_{C}^{*}(usp)\mathrm{r},$$\mathbb{C}:=H_{d}^{*}(cS, L_{Cus}^{2}p(\Gamma\backslash Gs)^{\infty})$

where $H_{d}^{*}$ is the differentiable cohomology and $L_{Cu}^{2}(sp\Gamma\backslash Gg)^{\infty}$ is the space

of smooth vectors in the space of square integrable functions generated

by cusp forms. As is shown in [BLS] this is a direct summand of the

usual $\mathrm{c}\mathrm{o}\mathrm{h}\mathrm{o}\mathrm{m}.0$logy of

$\Gamma$

.

The compatibility of base change with cohomology allows to prove

for any semi-simple simply connected split group $G$ a theorem proved

(5)

Theorem 3. Let $G$ be a simply connected semi-simple split group over

a totally real field $F_{0}$. $Co\mathrm{n}$sider a tower of cyclic $\mathrm{e}xt$ensions

$F_{0}.\subset F_{1}\subset\cdot X\cdot\subset F_{n}=E$ .

Let $S$ be a finite set of $pl\mathrm{a}$ces of $ECo\mathrm{n}\iota_{\mathrm{a}}i\mathrm{n}i\mathrm{n}g$ all archimed$\mathrm{e}\mathrm{a}npl\mathrm{a}$ces.

Then, any $S$-arithmetic subgroup $\Gamma$ of $G(E_{S})con$tains an S-arithmetic

subgroup $\Gamma’$ offinite index with

$\mathrm{n}$on trivial cuspidal cohomology:

$H_{cusp}^{*}(\Gamma’, \mathbb{C})\neq 0$

.

6 –About the proof

When $H$ is a unitary group, theorems similar to theorems 1 and 2 are

already in the literature (see for example [Clo2] and [Clo3]). The key

ingredients of the proofs, which rely on a trace formula comparison, are:

1 -the fundamental lenma for units in the base change situation [Kol];

2 –the extension of the fundamental lemma for base change to the full

unramified Hecke algebra using [Clol] (or [Labl])

3 –the stabilization of the trace formula using Euler-Poincare

func-tions at finite places following [Ko2].

Our proof follows the same familiar patterns. We observe that some

$\iota$

of the arguments in the papers quoted above are incomplete or

incor-rect. For example the proof ofthe $\mathrm{v}\mathrm{a}\dot{\mathrm{n}}\mathrm{i}_{\mathrm{S}}\mathrm{h}\mathrm{i}\mathrm{n}\mathrm{g}$ assertion in the

fundamental

lemma for elements that are not a norm is not correct in [Clol] (and the

same error is reproduced in [Labl]$)$. This can be corrected following a

suggestion of Kottwitz. In the proofofthe existence of lifting, as outlined

in [Clo3], it does not seem to be enough to deal with functions of regular

support and some control on the singular set seems necessary.

In [Lab2] we carry out the necessary work to correct and complete

these proofs and we extend the arguments to the general case. This

preprint is presently available under the internet address:

http:$\mathit{1}/\mathrm{w}\mathrm{w}\mathrm{w}.\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{h}.\mathrm{j}\mathrm{u}\mathrm{S}\mathrm{s}\mathrm{i}\mathrm{e}\mathrm{u}.\mathrm{f}\mathrm{r}/\sim 1\mathrm{a}\mathrm{b}\mathrm{e}\mathrm{s}\mathrm{S}\mathrm{e}/\mathrm{p}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{i}\mathrm{C}\mathrm{a}\mathrm{t}\mathrm{i}_{0}\mathrm{n}\mathrm{S}$

.

html REFERENCES

[AC] J. ARTHUR L. CLOZEL, Simple algebras, base change, and the

advanced theory of the trace formula, Annals of Math. Studies

(6)

[BLS] A. BOREL J.-P. LABESSE J. SCHWERMER, On the cuspidal

coho-mology of$S$-arithmetic subgroups of reductive groups overnumber

fields, Compositio Math. 102 (1996), pp. 1-40.

[Clol] L. CLOZEL, The fundamental lemma for $S\mathrm{t}\mathrm{a}b\iota_{eb\mathrm{e}}a\mathrm{S}$ chan

$ge$, Duke

Math. J. 61 No. 1 (1990), pp. 255-302.

[Clo2] L. CLOZEL, $r\mathrm{e}pre^{\nearrow}Sent\mathrm{a}tions$ galoisien$\mathrm{n}$es $as\mathrm{s}$oci\’ees a$ux$

repr\’eseDia-tion$s$ automorphes autoduales de $GL(n)$, Public. Math. IHES 73

(1991), pp. 97-145.

[Clo3] L. CLOZEL, On the cohomology ofKottwitz’s arithmetic varieties,

Duke Math. J. 72 (1993), pp. 757-795.

[DN] K. DOI H. NAGANUMA, On the functional equation of cert$\mathrm{a}i\mathrm{n}$

Dirichlet series, Invent. Math. 91969/1970 p.1-14

[Kol] R. KOTTWITZ, $B\mathrm{a}s\mathrm{e}$ change for units of Hecke

algebras,

Composi-tio Math. 60 (1986), pp.

237-250.

[Ko2] R. KOTTWITZ, Tamagawa numbers, Annals of Math. 127 (1988),

pp. $6\sim$29-646.

[Labl] J.-P. LABESSE, Le lemme$fo\mathrm{n}d$ament$\mathrm{a}l$pour le changemen$\mathrm{t}$ de $b\mathrm{a}\mathrm{s}e$

$st\mathrm{a}ble$, Duke Math. J. 61 No.2 (1990), pp. 519-530.

[Lab2] J.-P. LABESSE, Changem$\mathrm{e}nt$ de base $co\mathrm{n}$ditionn$el$pourles groupes

$re’ductif_{S}$, preprint (January 1998)

[Lan] $\mathrm{R}.\mathrm{P}$. LANGLANDS, $B$as

$\mathrm{e}$ Chan

$g\mathrm{e}$ for $GL(2)$, Annals of Math.

Stud-ies 96 Princeton Univ. Press (1980).

[Rog] J. ROGAWSKI, Automorphic Representations of Unitary Groups

in Three Variables, Annals of Math. Studies 123 Princeton Univ.

Press (1990).

[Sa] H. SAITO, $Au\mathrm{t}$omorphic forms and algebraic extensions of number

fields, Kyoto Univ., Lectures in Math. 8 Kinokuniya Book Store

Co. Ltd. Tokyo, Japan (1975).

Jean-Pierre LABESSE

Universit\’e Paris 7

Institut de Math\’ematiques de Jussieu (UMR 7586)

2 Place Jussieu

75005 Paris

参照

関連したドキュメント

Diaconu and Garrett [5,6] used a specific spectral identity to obtain sub- convex bounds for second moments of automorphic forms in GL(2) over any number field k.. That strategy

A structure theorem for ´etale morphisms (3.1.2) allows us to give a proof of the ´etale base change theorem following closely the proof in the rigid case.. We calculate the

– proper & smooth base change ← not the “point” of the proof – each commutative diagram → Ð ÐÐÐ... In some sense, the “point” of the proof was to establish the

Nevertheless, a dis- tributional Poincar´ e series may be constructed via an averaging map, and global automorphic Sobolev theory ensures the existence and uniqueness of an

Note that the Gysin isomorphism [20, Theorem 4.1.1] commutes with any base extension. The assertion follows from induction on the dimension of X by a similar method of Berthelot’s

We now show that the formation of the Harder-Narasimhan filtration commutes with base change, thus establishing the slope filtration theorem; the strategy is to show that a

We also show in 0.7 that Theorem 0.2 implies a new bound on the Fourier coefficients of automorphic functions in the case of nonuniform

In addition, we prove a (quasi-compact) base change theorem for rigid etale cohomology and a comparison theorem comparing rigid and algebraic etale cohomology of algebraic