In this thesis we prove that under certain conditions, if a special value of thesymmetric square L-function attached to a modular form f of appropriate weight, level and character vanish
Siegel and Klingen Eisenstein series
Siegel and Klingen Eisenstein series are induced from the maximal parabolic sub- groups P and Q of G = U(2,2) respectively (For the definitions of P and Q see section 2.2.) Hence by Remark 3.1.9, they are functions of one complex variable,
27 which we denote by s Let dp: P(A) > R4 be the modulus character of P(A),
(At)-} with A € Resxjq GL2(A), u € Up(A), and dg: Q(A) — RL be the modulus character of Q(A)
\L c a with ứ € ResxjgGli(A), [25] € U(1,1)(A) and u € Ug(A) As before, let
Ko = KossKo¿ denote the maximal compact subgroup of G(A) Using the Iwasawa decomposition G(A) = P(A)Ko we extend both characters dp and dg to functions on G(A) and denote these extensions again by ốp and dg It is easy to check that ôp(g) = 1 for g € P(A)N Ko and similarly dg(g) = 1 for g € Q(A)NKo, hence the ex- tensions are well-defined Moreover, it is clear that ốp (resp dg) is a P-automophic (resp Q-automorphic) form in the sense of Definition 3.1.1 with = 1 Fors ¢€C we have
(3.9) 6% € Ip(s—1/2,1) ={f: G(A) — Ơ | ƒ(muk) = ọp(m)(°~3)*3 ƒ(k), f is smooth and Ko-finite} and the local function ðb„ is the spherical vector for every place p of Q We similarly have dg € Ig(s — 1/2, 1).
P(Q)\G(Q) is called the Siegel Evsenstein series.
Remark 3.2.2 Note that Ep(g,s) = E(P,s— 1/2,1,ƒ,g), with E(P,s,1,f,g) as in Definition 3.1.7, where ƒ denotes the constant function with value 1.
The properties of #p(g, s) were investigated by Shimura in [Shi97] We summarize them in the following proposition.
Proposition 3.2.3 The series Ep(g,s) is absolutely convergent for Re(s) > 1 and can be meromorphically continued to the entire s-plane with only a simple pole at s = 1 Furthermore, the residue of Ep(g,s) at s = 1 is a constant function of g. The value of this constant is equal to
(3.10) ress) Ep(g, 3) = Aen where L{-,-) denotes the Dirichlet L-function.
Q(Q)\G(Q) is called the Klingen Eisenstein series.
Remark 3.2.5 In the case of Klingen Eisenstein series we also have Eg(g,s) E(Q,s—1/2,1, f,g), with E(Q,s,1, ƒ, g) as in Definition 3.1.7, where f denotes the constant function with value 1.
Proposition 3.2.6 The series Eg(g,s) converges absolutely for Re(s) > 1 and can be meromorphically continued to the entire s-plane The possible poles of Eo(g, s) are at most simple and are contained in the set {0,1/3, 2/3, 1}.
Proposition 3.2.6 follows from Lemma 1 in [RS91] The only difference is that instead of Eg(g,s), [RS91] uses an Eisenstein series that in section 4.3 we denote by E,(Z) The connection between Eg(g,s) and E,(Z) is provided by Lemma 4.3.3 In section 4.3 we will show that Eg(g,s) has a simple pole at s = 1 and calculate its residue.
Both Ep(g,s) and Eg(g,s) have their classical analogues, i.e., series in which g is replaced by a variable Z in the hermitian upper half-plane H Let g € G(R) be such that Z = gyi and set g = (gu, 1) € G(R) x G(A¿;) Define
We will show in Lemma 4.3.3 that
3eQ(2)\G(2) (in (1⁄))ss where for any 2 x 2 matrix M we denote its (i, j)-th entry by Ä;; It is also easy to show that
Ep(Z,s) = (detIm(yZ))*= ` — |det(c,Z+dy)|-*. yEP(Z)\G(Z) yEP(Z)\G(Z)Remark 3.2.7 Note that we use the same symbols #p(-, s) and Eg(-, s) to denote both the adelic and the classical Eisenstein series Which of the two kinds of Eisen- stein series are considered in a particular situation, can be recognized by looking at the argument (g or Z) We will continue this abuse of notation for other Eisenstein series we study.
Borel Eisenstein series 2 1 ee 30
In this section we give an example of an Eisenstein series induced from the Borel subgroup B of G As B is a minimal parabolic and G is of rank 2, such an Eisenstein series is a function of two complex variables which we denote by s and z To define it we will use the modulus functions ốp and dg defined in section 3.2.1 Note that B is contained in both P and Q, hence the following series
7€B(Q)\G(Q) is (at least formally) well defined Note that as the Levi subgroup of B is abelian (it is the torus 7'), the character 636% is a cusp form on T(A) The following proposition follows from [MW89], Proposition II.1.5.
Proposition 3.2.8 The series Eg(g,s,z) ts absolutely convergent for
It can be meromorphically continued to all of C x C and has poles along the lines z= 1/2 ands = 2/3.
Remark 3.2.9 It follows from the general theory (cf [Lan76], chapter 7) that by taking iterated residues of Eisenstein series induced from minimal parabolics one obtains Eisenstein series on other standard parabolics These series are usually referred to as residual Hisenstein series In fact Ep and Eg are residues of Hg taken with respect to the variable s and z respectively We will prove this fact in section4.3, but see also [Kim04], Remark 5.6.
Holomorphic Eisenstein series on SLz(2)
Siegel and Klingen Eisenstein series considered in section 3.2.1 are both defined using a section f which is the constant function with value 1 (cf Remarks 3.2.2 and bài
3.2.5) Classically this means that both series are of weight 0 and trivial character.
We will later need a variant of the Siegel Eisenstein series with positive weight and non-trivial character To motivate our definitions in section 3.2.4 we will now work out an example of the holomorphic Eisenstein series of weight m (for m an even positive integer) on SL.(Z) defined as
+eB;(Z)\SLa(Z) +€B›:(Z)\SLa(Z) Here z¡ denotes a variable in the complex upper half-plane H, By is the upper- triangular Borel subgroup of SLằ and j(y, 21) := Âyz1 + dự We will now redefine the series (3.11) in the context of Langlands theory developed in section 3.1 Let Kằ denote the maximal compact subgroup of SL2(A) with Ky = K2, K2+, where cos@ sing
—sing@ cos@ is the stabilizer of ¢ in SLo(R) Let By = T,U2 be the Levi decomposition of By with
1 Using the Iwasawa decomposition SL2(A) = B2{A) Ko we write g € SL2(A) as g = bk and define a Bj-automorphic form
#: Te(Q)U2(A) \ SLa(A) + C u(bk) = J (Koos 0) ”) where kyo € Koo is the co-component of k It is clear that y is well defined (i.e., trivial on Bo(R)M Koo) Let g = (goo, 1) € SLa(R) x SL¿(A;) with goo = dookoo € Bo(R) Koo Let 21 = 21 +0: := Goi € H Then boo yy”
Indeed, this follows from the fact that we can write ygu = bik, with
Let 62 : Bạ(A) — Rx denote the modulus character of SL2(A) defined by bp = lala * 2
As usual we extend 62 to a function on SLa¿(A) using the Iwasawa decomposition. Note that
We have € ùg,(—1/2, 1), with a local decomposition
Also 63 decomposes into a product of local factors in an obvious way Let f(g) := w(g)da(g)TM”.
We have f = Il, tp € Ip, (23, 1) It is clear that f, is spherical for p oo, but in contrast with examples in section 3.2.1, f, is not spherical In fact one has cos@ sing foo Joo = [lool Joo )e
—sing cosớ imo for g € SLe(R).
We define an adelic Eisenstein series by
7€B2(Q)\SL2(Q) Lemma 3.2.10 Write g = gqgooks € SL2(Q) SLo(R) Kor Then
Proof First note that the natural injection B,(Z) \ SLe(Z) — Bo(Q) \ SL2(Q) is a bijection, hence Em(g) = 3 ”„eứ;(zwsra(z) (Y9) It is enough to prove (3.14) for g = (Goo, 1) For such g, using (3.12), (3.13), and the definition of we get
Em(g) = Gasp jl Reo 1) = = Dean) “Tp aye 7 Foe) 7762.) Els) = J (Joo 1) (B00 2), Em (21):
Remark 3.2.11 In the notation of Definition 3.1.7 we have
Siegel Eisenstein series with positive weight
In this section we define a Siegel Eisenstein series with positive weight, level and non-trivial character For notation refer to chapter 2 Let m, N be integers with m >
0 and N > 0 Note that Ko is the stabilizer of iin G(R) Let ý: KX \ AZ — C* be a Hecke character of AX As explained in section 2.1, ~ has a local decomposition y= II, Wp, where p runs over all the places of Q Assume that and
Up(p) =1 iÍp # œ,z„ € OR,, and z,— 1€ NOKằ.
As before we set x = ]]jIyp Let dp denote the modulus character of P We define a P-automorphic form úp: Mp(Q)Up(A)\ G(A) + C by setting up(g)=0 ifg ¢ P(A)Ko(N) and yup (qh) = (det dạ)”túy (det dy) 2j(heo, i)" for ¢ € P(A), k € Ko(N).
Recall that for g € G(R), j(g, Z) := det(c,Z + đụ) Define a P-automorphic form f(g, 8) := up(g)Šp(g)°2.
We raise ốp to the power s/2 instead of s to make our notation consistent with [Shi97].
Note that f(g, 3) has a local decomposition f = |], fp, where ƒp(g, 8) = tps(8)ð/2(9)
(3.15) Hrp(qpEp) = 9 Wp(det dg,)'Up(detd.,) ifp| N,p # 00,
| Hoo (det dacs) i (Koo, i)7TM if p = œ and
(3.16) bpp uk | = | det Adet Ala,.A —
Thus fp € Ip((s — 1)/2,;ˆ) and if p{ Noo, fy is spherical.
Definition 3.2.12 The series yEP(Q)\G(Q) is called the Siegel Eisenstein series of weight m, level N and character w.
Remark 3.2.13 In the notation of Definition 3.1.7, we have
The series E(g,s, N,m,w) converges for Re (s) sufficiently large, and can be con- tinued to a meromorphic function on all of C (cf [Shi97], Proposition 19.1) It also has a complex analogue E(Z, s,m, w, N) defined by for Z = gol, 9 = 9Q9Gakt € G(Q)G(R)Kos(N) It follows from Lemma 18.7(3) of [Shi97] and formulas (16.40) and (16.48) of [Shi00] that
E(Z,s,m,, N) = ằ wy (det d,)~1 det(c,Z + d,)"TMx (3.17)
Tp (tUU, ww) by
(3.18) (Mov.9)0)(6) = | f0 hug) where ƒ € Ip-(v,w), and (U’)~ denotes the unipotent radical of the parabolic Pa, opposite to P’ The operator M(w,v,w) factors into a product of local intertwining operators defined by an obvious modification of (3.18).
From now on we restrict our discussion to the case when P’ = P, with 6 = A—{a} for some a € A is a maximal parabolic Then a}, c is one-dimensional and the map
37 đc — C defined by 2pp +> 1 (cf Example 3.1.4) identifies a1; with C In this case we write À/(u, s,) instead of M(w,2spp,) There exists a unique element wo € W with wo(a) < 0 such that wo(@) c A ([Kim04], section 4.3).
Definition 3.3.1 Let Q’ be a standard parabolic subgroup of G’ with Levi decom- position Q’ = Mg:Ug: Then the integral
Ugr(Q)\Ugi (A) is called the constant term of E(Po,s,w, f,g) along the parabolic Q.
Theorem 3.3.2 (Langlands, cf [Kim04], Theorem 5.3) Eg: (Po, s,0, ƒ, g) = 0 unless Q! = Pg or Q’ = Py ye) Let ƒ, be as in Definition 3.1.5 If Po # Pua), then
Remark 3.3.3 As asserted before, the Eisenstein series E( Po, s, ý, ƒ, g) has a mero- morphic continuation to the entire complex plane It is a theorem of Langlands that its poles are all simple and are the same as the poles of the various constant terms of E( Ps, s,v, f,g) In the next section we prove this fact for an Eisenstein series on the group U(1, 1) and calculate the residue explicitely.
The caseofU(1,l) 0.0.00 00000022 2G 37 4 The Petersson norm of a Maass lift 0 0.000 eee eee 43
Let B, denote the upper-triangular Borel subgroup of U(1,1) with Levi decom- position B, = TU), where
1 Let 6, : B,(A) — R, be the modulus character given by by u | = |aa|A qt for u € UẠ(A) Let K, = Ky denote the maximal compact subgroup of U(1, 1)(A) with a £6 =
F100 = € GLa(C) | |a|Ÿ + |B)? =1, ab ER being the maximal compact subgroup of U(1,1)(R) and Ais = ]],⁄„„ U(1,1)(2;).
As usually we extend 6, to a map on U(1,1)(A) using the Iwasawa decomposition. Set
Remark 3.3.4 The Eisenstein series (3.19) can also be written in the notation of Definition 3.1.7 In fact one has ụ‡ € Ip,(s — 1/2,1) and Euq,)(ứ,s) = E(ỡ,s— 1/2,1, f,g), where f is the constant function with value 1.
It again follows from the general theory of Langlands (but see also [Shi97] for a direct proof in our case) that the series Eu(ii(ứ, s) converges for Re(s) > 1 and continues meromorphically to the entire complex plane We now define a complex analogue of Eyiiiy(g,s) As SL2(R) acts transitively on H, so deos U(1,1)(R) D SL2(R) Hence for every z¡ € H there exists g € U(1,1)(R) such that z¡ = gol. Set g = (goo, 1) € U(1, 1)(R) x U(1, 1)(Ay).
Lemma 3.3.5 Let z, and g be as above Then 6;(g) = Im (2).
Proof Write go = [25] € U(1,1)(R) Then Im (z¡) = Im ((ai+b)(@+d)) |c + d|~? On the other hand [25] = XY, where cI2+ ldị2 |1 at+bd ore (|? + Ia?
Y =——- € te, vie?+Iđ? |e a after noting that |c? + d?| = |e|? + |d|? as cd € R Thus we have
Let z¡ and g be as in Lemma 3.3.5 Define the complex Eisenstein series corre- sponding to Eva,y (9; s) by
Lemma 3.3.5 together with the fact that the natural injection
By(Z) \ UG, 1)(Z) > By(Q) \ UMA, 1)(Q) is a bijection (cf [Shi97], Lemma 8.14) imply that
+€B:Z)\U4,1),) Let 2; = Re(z¡) and y, = Im(z) As Eua,p(2ìi + 1,5) = Eua,(2ì, s) the series Evy (415 3) possesses a Fourier expansion of the form
Arguing as in the proof of Theorem 1.6.1 in [Bum97] one sees that Ð ”„ o ca (ì, 5) for any fixed ¡ continues to a holomorphic function on the entire s-plane and for any fixed s decays exponentially as y; — oo In particular, for every 6 > 0, there exists a continuous function N(s) such that if y, > 4, then >? 49 |en(y1,8)| < N(s), where N(s) does not depend on y; Hence the poles of #u(¡,¡) (21, s) are the poles of co(yi, 3).
It is an easy exercise that co(yi,8) = Ep,(Bi,s — 1/2,1, f,g) with 2 = go2%, where
Ex, (Bi,s — 1/2,1, f,g) is the constant term of the series in Remark 3.3.4 Hence our conclusion is consistent with Remark 3.3.3.
Lemma 3.3.6 Let z‡ and g be as in Lemma 3.3.5 Then ¢(2s — 1) P (s — 5)
Note that in our case B; = Py is both a minimal and a maximal parabolic and we denote by a the positive root determined by Bj, ie.,
We have pz, = 5a Since the Weyl group of U(1, 1) has only two elements, it is clear that wo = [, T1], where wo is the element defined in section 3.3.1 and B, = Py Pup(@) a8 ệ = 0 Hence by Theorem 3.3.2 we have
We will now calculate M(wo, s,1)ð‡*1?, The operator M(wo,s,1) factors into a product of local intertwining operators
(plore 8.19889") (op) = fh đi)" (and,
=—1 Ấp for u € U;(Q,) It is easy to see (cf e.g., [Tan99], section 2) that
(3.27) Mp(wo, s, Lối, ” = Tope jp for p # oo For p = co we have
Note that since 61,.o(tuk) = 61, (t) for t € 71(R), u € U,(R) and k € Ki we can assume without loss of generality that 9 = [“g-:| € Ti(R) Then
Using formula (3.21) we see that
—1/ |1 2 1 la|~? ổn co Joo = ~ “Mh a+ ]sí8)-P 14 [ae as x € R Thus oo s+1/2 s+1/2 — |l„|—2s=1 !
(3.29) (Mao(wo,s, 1)6152/7) (Goo) = lal / (mm) "
Making a change of variable z +> |a| ”z we see that the right-hand side of (3.29) equals cai (P _ - T)2s+1 1 2\—s 1/2 =ỗ ss(0œ s+1/2 o9 [bate = blo) VE êm
Hence the above calculations together with (3.25), (3.26), (3.27), and (3.28) give
By (Bu,s,1, f9) = bilo) + VR EAL ĐÓ 5 (gyro (28+ 1)T (s+)
Using formula (3.24) we obtain the lemma L]
Corollary 3.3.7 The series Euq (2z, s) has a simple pole at s = 1 with residue the constant function with value 3.
Proof As remarked before Lemma 3.3.6, the difference Eyc1,1)(21, 8) — €o(ì, 8) con- tinues to a holomorphic function on the entire s-plane Hence we have
The Petersson norm of a Maass lift
The goal of this chapter is to express the denominator of Cp, in formula (1.2) by an L-function of f We start out by defining the lifting procedure (section 4.1) and then carry out the calculation of the Petersson norm (#;, Fy) in the following sections.
Maass lifts 2 Q nu ng g g v v v v v kg kg va 43
Let H, as before, denote the complex upper half-plane The space H x C x C affords an action of the Jacobi modular group IY := SL2(Z) x Ó?, under which ([¢5],,) takes (7,z,w) €Hx Cx C to (44, 4), 4).
@:HxCxC3C is called a Jacobi form of weight k and index m if for every [2°] € SL¿(Z) and a b czw ar+6 # 10
P= ðlum c d = (er+4) -{ me) Pm (3555) and ở = ó|m|À, H] := e(mAA + Az + AW) bm(T, z + AT + pw + À£ + fi).
Let k be a positive integer divisible by 4 We denote by N the set of positive integers and put No := NU {0} Let F be a hermitian cusp form of weight k and full level By rearranging the Fourier expansion of F we obtain
(4.1) F(Z) = ` bm(r, 2, w)e(mr*) meN where Z = [7,2] € H and
1ENo t€ZOK tom tt fo(r) gives an injection of J¿¡, the space of Jacobi forms of weight k and index 1, into S;,_1 (4,(=*)) If we put = ¢; and define ó by Ớ|¿-i[¿ `] = fo, the composite F ++ ¢,(7,2,w) + fo(7) > ¢ gives the isomor- phism alluded to in Theorem 4.1.3 We will refer to it as the Maass correspon- dence Denoting this isomorphism by 2, we can map any normalized Hecke eigenform f = Sasi b(n)q” € Ses (4, (=4)) to the element Fy := Q~!(ƒ — f?) € Sÿ22%(T7).
Here f? = ons O(n)qQ” The Maass correspondence is Hecke equivariant in a sense, which will be explained in section 5.2.4 Note that Fy = —Fyằ and Fy # 0 if and only if ƒ # f?.
Definition 4.1.4 If ƒ # f?, then Fy 4 0 is called the Maass lift of f or the CAP lift of ƒ.
Proposition 4.1.5 If f = 30,5, b(n)4" € Sp_1 (4, (=4)) is a normalized eigenform, then say 6(n) — bí) iÊn #1 (mod 4)
Proof This follows from formula (4) on page 670 in [Kri91] O
The Petersson normof Fp Q Q HQ HQ HQ ee, 45
Our goal in this section is to express the inner product (F;, Fy) in terms ofL-functions associated to f Let F(Z) = 3 m>o ÓmÍT, z,0)e(mT7”) and F(Z) mee Um(T, Z, w)e(mr*) be the Fourier-Jacobi expansions of two hermitian modular forms F and F” of weight k and full level Raghavan and Sengupta [RS91] consider the following Dirichlet series:
Drr(s) = Cx(s —k+ 3)c(2s — 2k+ 4) ằ (Om; Um) m Ÿ, m>0 where for two Jacobi forms ¡ and ta of weight k and index m, we put k —xm|z—1ứ12
(Úi, 92) = vr(7, 2, +U)2(T, Z, w)v € i du, gs with 7 = utiv, Z = Zo + 021, W = Wot 000, U,V, 2%, 21,Wo,Wi € R and du v~‘du do dzp dz; dứa dw; The integration is over a fundamental domain Z” for the action of TY on H x C x C defined in section 4.1 Here ¢(s) denotes the Riemann zeta function and ¢x(s) the Dedekind zeta function of K.
Theorem 4.2.1 (Raghavan, Sengupta, [RS91], Theorem 1) The Dirichlet series Dr p(s) can be meromorphically continued to the entire s-plane The function
Dp p(s) = (4n) ”°TÍ(3)I(s — k + 2)P(s — k + 3)Dr,r(s) is holomorphic in s except for possible simple poles ats = k— 3, k— 2, k— 1, k and satisfies a functional equation Dy p(s) = Dip (2k — 8 — 3).
To derive the desired formula for (Fr, Fr) we will use an identity proved in [RS91] that involves a variant of the Klingen Eisenstein series, which we now define The relevant properties of the Klingen Eisenstein series were discussed in section 3.2.1.
For a 2 x 2 matrix M, denote by Ä/;; the (2, 7)-th entry of M Let C be the subgroup of ['z consisting of all matrices whose last row is 0 0 0 1) We will later relate
C to the Klingen parabolic subgroup of G Set
The series converges for Re (s) > 3 ([RS91], Lemma 1) In the course of the proof of Theorem 4.2.1 Raghavan and Sengupta establish the following identity
They also prove that E*(Z) can be analytically continued in s to the entire complex plane except for possible simple poles at s = 0, 1,2, 3.
The goal of the next two sections is to prove the following theorem:
Theorem 4.2.2 The function E* has a simple pole at 3 with residue (independent of Z) equal to +c@).
Assuming Theorem 4.2.2, we can relate (F;,F;) to the symmetric square L- function L(Symm? f,s) of f using the following formula due to Gritsenko [Gri90al:
Here we define L(Symm? ƒ, s) for a normalized eigenform f = 5>°2, a(n)g” as an n=l Euler product:
_ —g —sv1—1 x I] [q — 21p *)(1 = apiapap ` ®)(1 — Q2 2p )| pz⁄2 where the complex numbers a, and apằ are defined by the equation
By Fact 2.3.1, we have a(p) = (=) a(p) for p+ 2 This implies that a(p) € R if p splits in K and a(p) € iR if p is inert in K It follows that @j7 = ap, for split p and đp1 = —Qp.2 for inert p We will call ay and œ; › the classical p-Satake parameters of f to distinguish them from the “representation-theoretic” Satake parameters which we will define in chapter 9.
Combining formulas (4.5) and (4.7) with Theorem 4.2.2 we obtain:
Finally, to relate (ở, 61) to (f, f), in section 4.5 we will prove the following lemma.
Lemma 4.2.3 The following identity holds:
Combining Lemma 4.2.3 with formula (4.9) we finally obtain:
Theorem 4.2.4 The following identity holds:
Residue of the Klingen Eisenstein §@Ties 0.2 eee eens 48
This section will be devoted to proving Theorem 4.2.2.
Proof of Theorem 4.2.2 We will now calculate the residue of the Klingen Eisenstein series defined in section 3.2.1, and at the end relate it to the residue of E,(Z) Once we have the residue of F(Z), the residue of E(Z) can be determined by the formula (4.6) For notation refer to chapter 2.
Let Ko = Kooko, be the maximal compact subgroup of G(A) First note that
Bc Q where B = TUg denotes the Borel subrgoup of G and Q = MgUg the Klingen parabolic This embedding on the level of Levi subgroups and unipotent radicals is explicitely given by:
We now define four characters on Mp(A) and Mg(A) The first two are: the modulus characters ốp and dg defined by formulas (3.7) and (3.8) respectively The other two are defined as follows: Let l “ , € Mo(A) Since [25] € U(,1)(A), we can use the Iwasawa decomposition for U(1,1)(A) with respect to the upper- triangular Borel to write [¢5] = [* 2,]k with k € Ki, where Kj is as in section
3.3.2 Note that if k = [i 2 I, then tr BỊ € Ko We define 1 kạ ka by
1 aol Finally the fourth character gp: Mp(A) — Ry is defined by:
DL * ky ko y ks kg _ù—— op = |zy (zy )la, zt kị ks xa ký ki | j where we used the Iwasawa decomposition for GL2(Ax) = Resx/q GL›(A) with respect to its upper-triangular Borel Br, and its maximal compact subgroup KR U(2) [Trico GL2(Ox,) to write A € GLo(Ax) as ol ki ko
We again have Ra ka kk | € Ko. kl, ki,
It is easy to check that all four characters can be extended to the adelic points of the corresponding parabolics by setting them to be trivial on the parts of the unipotent radicals Then using the Iwasawa decompositions
(4.16) G(A) = B(A) Ko = P(A) Ko = Q(A) Ko the characters can be extended to functions on G(A) One also verifies that
(4.17) 535% = 6o o = 64740) for any complex numbers s and z Let
+ 2/3 and Re (2z) > 1/2 and admits meromorphic continuation to all of C2 Using identity (4.17) and rearranging terms we get:
Es(g.s,z):= ` ôq(xd) "7 ` da(eaxg)”= 2 2
+eP(Q)\G(Q) aeB(Q)\P(Q) Let Eu(,;(ứ, s) be the Eisenstein series on U(1,1) defined by formula (3.19) We also define an Eisenstein series on Resx/q GL2(A) by:
+€Bn(Q)\Resx/a GLa(Q) where dp denotes the modulus character on Bp defined by:
On the A-points we can extend 7g to a map G(A) — U(1,1)(A)/K¡ and mp toa map G(A) — Resx/q GL2(A)/Ke by declaring them to be trivial on Ko Hence we can rewrite formulas (4.19) as
(4.24) Es(g.sz):= ` ụa(xứ)°*fEvu(ma(xe).2z) =
As remarked in section 3.3.2 the series Eyii1)(g,s8) is absolutely convergent for Re(s) > 1 and admits meromorphic continuation to the entire s-plane By Corollary 3.3.7 it has a simple pole at s = 1 with residue `,
Proposition 4.3.1 (i) For any fixed s € C with Re (5) > 2/3 the function Ep(g, s, z) has a simple pole at z = 1/2 and
3 where Eo(g, s} is the adelic Klingen Eisenstein series (cf Definition 3.2.4).
(ii) For any fixed z € C with Re(z) > 1/2 the function Ep(g, s, z) has a simple pole at s = 2/3 and
(4.26) TeS,_2 B(g, 8,2) 602) 7 (0 5 + :) where Ep (g, z) is the adelic Siegel Eisenstein series (cf Definition 3.2.1).
We will prove this proposition in section 4.4.
We are now going to show that Eg(g,s) has a pole at s = 1, determine its residue and relate Eo(ứ, s) to E;(Z), which will allow us to calculate the residue of the latter.
Lemma 4.3.2 The Eisenstein series ERsy,o GL2 (g,s) is absolutely convergent for Re(s) > 1, admits meromorphic continuation to the entire s-plane and has a simple pole at s = 1 with residue z2
Proof The proof is very similar to the proof of the analogous statement in the case of the group U(1,1), so we omit it here The latter proof has been carried out in detail in section 3.3.2 O
Using identity (4.24) together with Proposition 4.3.1 and interchanging the order in which we take residues we obtain:
(4.29) res, _ and thus we finally get
All that remains now is to relate Eo(ứ, s) to E;(2).
Lemma 4.3.3 Let g = (goo,1) € G(A) and Z = gx„ì Then
Proof First note that det Im yZ )
E,(Z)=4 ——=—]›:4) > (Gmtin where C’ is the subgroup of Ứz consisting of matrices whose last row is of the form
0 0 0 a with a € OF Moreover we have C’ = wQzw7! with tu = | and 1
——_—_ = = = yeC\Pz (Im +Z2)in +eQz\Tz (Im 0u Z)in
+€Qz\Tz (Im yw! Z) 2.2 +€Q9z\Tz (Im 1)ằ2 as tu € Ủy.
Now for y € Tz we have ổo(y9) = ôổo(qg), where g = (goo, 1) and Goo = qookoo c with do € Q(R), ko € Kow If do = um with m = _— i € Mg(R) and d u € Ug(R), then ðo(9) = 5q(um) = ôq(m(m~”um)) = ôq(m) = |aj.
A direct calculation shows that det Im u(mi) = detImmi and that (Im u(mi))o2 (Im mi)o2 On the other hand
Immi ee _ (ci+d)(ci-+d) hence we have det Im yZ
The lemma now follows from the fact that the natural injection
Qz \Tz — Q(Q) \ G(Q) is a bijection This is a consequence of the identity Q(A) = Q(Q) Q(R) Q([ptoo 2). which follows from Lemma 8.14 of [Shi97] O
Using Lemma 4.3.3 and formula (4.30) we conclude that E,(Z) has a simple pole( at s = 3 with residue - on mat {= `) and hence that E7(Z) has a simple pole at )L(3,(=) ) you) - +,¢(3) This finishes the proof of Theorem
Eo(g,s) as a residual Eisenstein series 0-00 200004 56
This section is devoted to proving Proposition 4.3.1 We will only present a proof of part (i) of the proposition as the proof of (ii) is completely analogous In what follows Z will denote a variable in the hermitian upper half-plane H, and z¡ a variable in the complex upper half-plane H Otherwise we use notation from the previous sections As before we denote by Ko the maximal compact subgroup of G(A) Write g = ứqứk € G(A) with ga € G(Q), go € G(R) and k € Ky We have Èp(g,s,z) = Ep(go,8,2) and Eo(g,s) = Eg(go,8), hence it is enough to prove (4.25) for g = (ứứs, 1) € G(R) x G(A¿;) Let Ay denote the maximal compact subgroup of U(1,1)(A) and let mg : G(A) — U(1,1)(A)/K be as in formula (4.22).
Lemma 4.4.1 If g = (goo, 1) € G(A), then Im (zo(g)2) = Im (g+Ì)2a.
Remark 4.4.2 Note that for any 2 x 2 matrix M with entries in C one has
Im (M22) = (Im(M))o2 Hence the conclusion of Lemma 4.4.1 can also be writ- ten as Im (ro(g)?) = Im ((gooi)2.2) (cf the proof of Lemma 4.3.3).
Proof of Lemma 4.4.1 Using the Iwasawa decomposition for G(R) we can write
Jo = umk € Ug(R)MQ(R)Ko,00 with m = _ ) As Ua(R) Mg(R) = Mo(R) Ue(R) we have 79(g)i = (ai + e d b)(ci + d)~", hence Im (779(g)i) = |ei + d|-?.
On the other hand gi = umki = umi A direct calculation now shows that
(Im (umi))o2 = (Im (mi))22 = |ci + dị”.
Lemma 4.4.3 For any Z € H, there exists y € Q(Z) such that (Im7Z)oo > š. hư
Proof Since SL2(Z) C U(1,1)(Z), for every Z € H there exists +' € U(1,1)(Z) such that 7'(Z.2) belongs to the standard fundamental domain of SL2(Z) on the complex upper half-plane H In partcular Im(7(Zo2)) > 3 Suppose 7 = [5] Then 2 d
+:= | a, | € Q(Z) and a simple calculation shows that 1 d e
The lemma follows by Remark 4.4.2 LÌ
Lemma 4.4.4 For every Z € ?(, we have sup Im(yZ) < co.
+€Tz Proof This is just an easy adaptation to the case of hermitian modular forms of the proof of Hilfsatz 2.10 of [Fre83] L1
Proposition 4.4.5 Let 6 > 0 and g = (0+, 1) € G(R) x G(A¿) For everys EC with Re(s) > 1 + ỗ and every z € C with |z — 3| < 6, the series
(4.33) D:=|z-1/2| SY) lóa(xứ)**”8Euu(na(9).22)|
Proof It follows from [Shi97], Lemma 8.14 that the natural injection
Q(Z) \ G(Z) = Q(Q)\ G(Q) is a bijection Hence in (4.33) we can replace >) cgay c(qy With dD yeq¢z)\ cz): Moreover, as shown in the proof of Lemma 4.3.3 det Im(yZ) _ 1/3 dim Z))aa 0909) where Z = gai, y € G(Z) For g’ = (7,1) € U(1,1)(R) x U(1,1)(A¿) let z¡ Joot € H For z’ € C define a complex Eisenstein series by setting
The series EZu(i,+)(2¡, 2’) was investigated in section 3.3.2 where we showed that
In our present case (4.34) leads to
Eua,b(ma(39).2z) = Eu (te(Vg) ot, 22).
Hence using the same arguments as in the proof of Lemma 4.3.3 we see that b= X (qaieaZ)\""
As 9 = (Goo, 1) and y € G(Z) C Ko, we have 7o(9)s = o((Yứỉs;1))s By
Lemmas 4.4.1 and 4.4.3 we can find a set S of representatives of Q(Z) \ G(Z) such that for every y € Š we have
As discussed in section 3.3.2 the series Ey(i,1)(21, 22) has a Fourier expansion of the form
Eva, (21,22) = › e„(2z, Im (z 2minRe (1) neZ and Ey1,1)(21,2z) — co(2z,Im (z¡)) for every fixed z¡ continues to a holomorphic
? function on the entire z-plane and for every fixed z is rapidly decreasing as Im (z¡) — oo It follows that for any given N > 0 there exists a constant M(N) (independent of z¡ and independent of z as long as |z ~ 1/2] < ở) such that |Euqa,(2ì, 22) — co(2z, Im (z¡))| < M(N) as long as Im (z¡) > N Set
Taking N = 1/2, we see by formula (4.35) that there exists a constant M (indepen- dent of +) such that | Bui) (ty + iyy,2z)| < M + |co(2z, yy)| By Lemmas 3.3.5 and 3.3.6 there exists a positive constant C' independent of z and of + such that
|z — 1/2|lco(2z, yy)| < dys? + Cy, for jz — 1/2| < 6 Since y, > 1/2 and |z ~ 1/2} < 6 we have two cases: either 1/2 < yy < l or y, > 1 In the first case |z — 1/2||co(2z, x)| < C’ for some positive constant C’ and in the second case |z — 1/2||co(2z, wx„)| < |y,|!*?° Hence in either case we have
Thus we conclude that there exists a positive constant A such that
For s’ € C lying inside the region of absolute convergence of F(Z) let det Im (yZ) \*
|E|e(Z) := ằ (= ae) yEQ(Z)\G(Z) denote the majorant of #;(Z) By formula (4.36) we have det Im (y2) Sete
Note that |E|3,42.(Z) is well-defined (i.e 3s + 2z is in the region of absolute con- vergence of #„(Z)) by our assumption on s and z Denote the second term of the right-hand side of formula (4.37) by Do Then det Im (+Z))!??.
By Lemma 4.4.4 there exists a constant M(Z) such that detIm(yZ) < M(Z) and hence
Dạ < AM(Z)'*” | E]gs492-(1428) < 00 as Re (3s + 2z — (1 + 26)) > 3 by our assumptions on z and s This finishes the proof L]
Proof of Proposition 4.3.1 We need to show that for a fixed s € C with Re (s) > 2/3 and for every € > 0 there exists ở > 0 such that |z — 1/2| < 6 implies
As remarked at the beginning of the section we can assume without loss of gen- erality that g = (go, 1) € G(R) x G(A¿) We first show that (4.38) holds for s with Re(s) > 1 Fix s € C with Re(s) > 1 and 6’ > 0 such that 0 < ở < Re(s) — 1. From now on assume |z — 1/2] < ở Fix a set S of representatives of Q(Q) \ G(Q).
By Proposition 4.4.5 and the fact that Eo(g,s’) converges absolutely for s with
Re (3) > 1, there exists a finite subset S; of S such that the following two inequali- ties:
Z— 1 | ụa(%ứ9)°†?3Euu,(a(ứ),2z)| < mo are simultaneously satsfied Here Sz denotes the complement of S, in S We have D(z) < D,(z) + Do(z), where
D;(z) := (: — 5) À ` ụo(xứ)°”?!8Eva(v(39),2z) — 5 À ` ụo(xg)°?18|,
Note that if we replace 6’ with a smaller 6” > 0, then estimates (4.39) and (4.40) remain true as long as |z — 1/2] < 6” for the same choice of S; Hence we find 6 > 0 with 6 < ở such that D¡(z) < § This is clearly possible as D,(z) is a finite sum and 3/27 is the residue of EZu(q,(mo(ứ), 2z) at z = 1/2 by Corollary 3.3.7 On the other hand Do(z) < Ds(z) + Da(z), where
D3(z) = oo ủ | ụa(+9)°°?'3Euaa(ma(x3).22)| 1 and
D„(z) := = ằ |(đa(sứ))°?%I : yES2 Formulas (4.39) and (4.40) imply now that D3(z) < e/4 and ¿(z) < €/4 Hence
D(z) < Dy(z) + Do(z) < Di(z) + D3(z) + Da(z) < < as desired.
We have thus established the equality resz~1/2#/g(g, s, z) = 2 EQ(9, s+ 1/3) for s with Re(s) > 1 However, both sides are meromorphic functions in s and since the right-hand side is holomorphic for Re(s) > 2/3, so must be the left-hand side.Hence they agree for Re(s) > 2/3 O
Inner product formula for Jacobi forms 00 0002 ee eae 61 5 Hecke operators 1 1 HQ ng Quà ng kg gà gi kg sa 65
This section is devoted to proving Lemma 4.2.3.
Proof of Lemma 4.2.8 Let \ and w2 denote two Jacobi forms of weight k and index m It is easy to show that
(4.41) (wr, ve) = [ *( w1(7, z, 10)2(T, Z, we = dz dz, đua dus) du dù,
F Fy where F is the standard fundamental domain for the action of SL2(Z) on the complex upper half-plane and Z; C {7} x C x C is a fundamental domain for the action of the matrices F 0 | and h d ul (A, € Ox) on C x C After performing a change of variables on C x C (keeping 7 fixed) zZ=z+U w=z-u, and denoting by Z7; the fundamental domain Z7; in the new variables, the integral over F, in (4.41) becomes
8 Je tì, 2’, w')pa(7, 2/, w’)ore D dz dz, dwy dw max.
Set v1 = wo = ị, where ở is the first Fourier-Jacobi coefficient of the CAP form F,; Using formula (4.3) we can write:
(4.42) (di, 1) = sob [ae )f#u(r)u~*®1(t,,T) du du teA cA with
(4.43) Fr `aet+O bet’ +OxK xe 9 (lm (Re wy) đạo dz duc dwy.
Changing variables again we get
(4.44) I(t,t.r)= So Soe N(d)7) Ly Ip a€t+OxK bet’ +OK with h= | e(2x’Re (a) — 22’Re(b)) e52*1) đa? dri,
Hi where (] is the parallelogram in C spanned by the two R-linearly independent complex numbers 1 and 7, and zÍ = zạ +Zz¡ € C, with 2,7, € R Before we define
Iz we note that J; can be written as
(4.45) = | cv 41) (| e(2z “Re (a) — 2zRe (b)) ax, dr’.
Now the integral inside the parantheses in (4.45) equals e-#”Re (9): if Re (a) = Re (b) and 0 otherwise Hence
4 [im c~45@¡)? e-87Re (4) #1 de! if Re (a) = Re (b)
0 if Re (a) # Re (0) The integral
Q2 where {22 denotes the region in the complex plane spanned by the two R-linearly independent complex numbers 1 and —7 and y’ = ¿ + iy, € C with yg, y, € R, can be handled in a similar way In fact one gets:
0 if Im (a) # Im (6) Substituting (4.46) and (4.47) into (4.44) one sees that /(f,f,7) = 0 if t # t’, and that after rearranging terms
(4.49) (ởì, dị) = / ` #ứ )#,(} U*~* du du du.
F teA From this it follows that 7,0, N >0 be integers Let ý : A} — CX be a Hecke character, which we think of as a char- acter : (Resx/q GL) (A) — CX We have a factorization = I], tứ; into local characters Wp : (Resx/q GL¡) (Q,) + CX Assume that w is such that and
Up(p)=1 ifp#o, zp€ểX,, andz,—1€ NOxằ.
As before we put Yy := [],)1 Up We will sometimes abuse notation and write wy for the character qy : (Z/NZ)* — C* which makes the diagram
Recall that in section 2.2 we defined
TẠ(N):= cTz|A,B,De Mạ(Oy), C € M;(NOy)A B
C D Denote by M=>(N,w) the space of functions F : H — C such that
F\|my = wn (det a,)F = wn (det d,) 1 F for all y € 2(N) Note that we do not require F to be holomorphic.
Example 5.2.1 Let (2, s,m, ú, N) be as in section 3.2.4 Then E(Z,s,m,w,N) €
We now proceed as in section 5.1 For details, see [Gri90a] Let
For a € Ab(N), the double coset space Tầ(N)aIÿ(N) decomposes into a finite disjoint union of right cosets
ThN)ar BN) = ]]Tš(09)4, with a; € AR(N) One could define R'(N) as the Z-module generated by the double cosets Tộ(V)a[(N), but this does not give the correct Z-module structure on the Hecke algebra We will return to this issue later Instead we define R®(N) to be the C-vector space generated by distinct double cosets [2(N)aI2(N) There exists a C-linear map
Rh(N) = Endc(MẸP(N,9)) given by
(Fầ(M)aTi(M)]|Ƒ = F|m[T8(N)aT8(CM)| = À ` bw (det das) F|maj. j
We denote the image of this map by T®(m, N, w).
Definition 5.2.2 A function F € M2®(N, 0) is called a Hecke eigenform or simply an eigenform if it is a simultaneous eigenfunction for all T € Thứn, N,v).
For a rational prime p we define operators
Lemma 5.2.3 Ifp{N we have the following decompositions:
(5.7) tt xuðeZ/p pả bEOK /p h ploy h KT
+€Z/p - 1 deOx /p " P yEZ/p for any prime p, re=tyn) | r, | TRÚM 1 la ỉ8 * T Trụ
58) = JJ ủnm|?? "fu Tp nw[ 7l o,Ô,+€Z/p TTA TS ãccZ/p P
U H TẠ(N) ng LITR(M) | ": $€Z/p 7 for p= TT split or ramified in K and
(5.9) Pl œéOy/p -@1 la B PS PỊ
U JT | ? 2 Ju Jf n0) | 7| œ€Okr/p —ðp p +eOkyjp P
BEZ/p? EZ /p for p inert in K.
Proof For the decomposition of 7” see [Kri91], section 7 The decompositions of 7? and of T?, can be obtained in a similar way We omit the calculations.
We now describe the action of the operators T?, TÌ, and T® on the Fourier coefficients of hermitian modular forms From now on we deal exclusively with holomorphic forms.
Lemma 5.2.4 Let FE Mn(N,w) with Fourier expansion
(5.10) crpr(B) =p" “*cr(p *B) + (Np)p ”ep(pB)+
3 | Q — where Np denotes the norm of the ideal (p) Moreover, for p = 1T split or ramified ink, pt N,
TT T7 ocZ/p Hi Wœ 7 and for p inert in K, pt N,
Dp 1 aeOry/p —a@ Ì Pp Proof This follows directly from Lemma 5.2.3 L]
Remark 5.2.5 Note that it follows from Lemma 5.2.4, that the Fourier coefficients of T?F and of TF for p split or ramified need not be algebraic integers even if the
Definition 5.2.6 Whenever the weight m is understood from the context, we set T® to be the image of T"(m,1,1) inside the space of C-linear endomorphisms of Sm(Tz), the (holomorphic) hermitian cusp forms of weight m and level 1.
Proposition 5.2.7 (Gritsenko, [Gri90a]) For any split or ramified prime p = ni set) := {12, Tệ, Tạ} and for any inert prime p, set Xi, := {Tp,TP,} The Hecke algebra T® is generated as a C-algebra by the set U, =p.
Satake parametrization 2 2 2 ee ee 74
From now on we fix m and assume that N = 1 We will use the Satake homomor- phism (1.e., the map Sat, defined below) to realize T® as a subring of a polynomial ring over C This leads to a notion of classical Satake parameters We emphasize the word classical to distinguish them from the representation-theoretic Satake pa- rameters which will be introduced in Chapter 9 For details we refer the reader to[Gri90b] and [HS83].
First let p be a prime inert or ramified in kK We define the Hecke algebra TEằ for G,,(Z,) to be the C-vector space consisting of finite sums }7 caG,(Zp)aG,(Z,p) with cy € C and the usual law of multiplication (cf [HS83]) We define a C-linear map
Sat;: TÔ „ => CÍZo,1,2,g ,21 ,Z2 | as follows: write Œ„(2g)aG,(Z„) as a disjoint union of right cosets ||, G,(Zp)a; with p Thẻ pes
Satp(G,(Zp)aG.(Zp)) = ằ xq” (é ứ\) “2 (p 2;)°22
(cf [HS83]) Every eigenform F € S,(I'z) induces a C-linear map Arc : Tệ„ > C sending to T € TEằ to the eigenvalue of T corresponding to F’ This map extends to a C-linear map
The complex numbers À;;(F) := Àpò(Z;), 7 = 0,1,2 are called the classical p- Satake parameters of F If F is clear from the context we sometimes write A); instead of Ap j(F).
Now consider a split prime p = 77 Let 7, and ig denote the two different embeddings of K into Q, We identify G,,(Q,) with the group Go = {(91,92) € GLi(Q,) x GLI4(Qp)|lứJứ) = ủ((:.92))J} via the map a @ 1 + (4:(a), (a), a € K We then identify the group Go with the group GL4(Q,) x GL1(Q,) via
(91,92) > (91, 0((ứỡ, 92))) Similarly G(Q,) is identified with GL4(Q,) (It is con- venient to introduce a change of variable, which makes the image of G,,(Q,) in GL4(Q,) x GL4(Q,) take the form (91, ((ỉi, 92))(g{)~") (cf [Gri90b}).
In this case the Hecke algebra TZ, , of G„(Q;) = GL4(Q,) x GL¡(Q;) is generated by double cosets Tal, with a = (ai,ứs) € GL4(Q,) x GLi(Q,) and ẽ' = ẽỡ xT, with [; = GL4(Z,) and T, = GL¡(Z¿) Inside TỀ,, is the Hecke algebra TG, for G(Q,) = GL4(Q,) generated by double cosets [ứĂ['; In fact we may identify
TEằ with a polynomial ring over TS, in indeterminates zp and zp‘, by mapping
TY, x Pe(ai, @2)Ty x Dạ = (TiaiP,)(T2a2F2) to the element TiaTirp re),
On the other hand there is a ring injection TS, — Clx1, v2, 23,24] defined by Daly = [[Tib¿ ye, Ti (p7'21)*3, where 8; € GL4(Q,) has the form pe * * * | pers * * ps
As before each eigenform F € S,,(I'z) determines a set of classical p-Satake parameters of F which we denote by Apj, 7 = 0, ,4.
Remark 5.2.8 A similar description holds for the elliptic Hecke algebra and one obtains in this way (for every p not dividing the level of the elliptic eigenform in question) the classical Satake parameters, which we denote by œ„ and apằ They agree with the Satake parameters defined in section 4.2 We omit the details.
Integral structure of the hermitian Hecke algebra
From now on we take m = k, a positive integer divisible by 4 We now define an integral structure on Th, Let F € S,(['z) be an eigenform For every rational prime
77 p, let Àp;(#') denote the classical p-Satake parameters of F’ Let p be a prime of Ox lying over p Set
Apu(F) = (Np) OP a" (ppg (F), where w is the unique Hecke character of K unramified at all finite places with infinity type Woo(Zoo) = 2)”, See also Remark 6.2.4 Zz
Definition 5.2.9 The elements Aằ;(F) will be called the Galois-Satake parameters of F atp.
By Theorem 10.1.3 (which will be discussed in section 10) to every eigenform
F € &(Tz), one can attach a Galois representation pr: Gx — GL,(Q,), such that pr is unramified at primes p which do not divide 2ý and for such p the set of eigenvalues of pr(Frob,) coincides with the set of Galois-Satake parameters of F at p Let órg(X) be the characteristic polynomial of pr(Froby,).
The coefficients a; can be determined from the explicit form of the map Sat, (cf. [Gri90a], section 2) One has
Ga = ],where Àrc(7) denotes the eigenvalue of T corresponding to the eigenform #'.
Proposition 5.2.11 The space Š„(Úz) has a basis consisting of eigenforms.
Proof This is a standard argument, which uses the fact that T® is commutative and all T € T® are self-adjoint L]
Example 5.2.10 and Proposition 5.2.11 motivate the following definition of the integral structure on T® For a split or ramified prime p = Z7 set and for an inert prime p set up := {77T}.
Definition 5.2.12 We set Tỷ (resp ToT) to be the Z-subalgebra of T” generated by U, 2p (respectively by U4 Up) For any Z-algebra A, we set Th := Tỳ @z A and Tt := Tht @; A.
Note that by Proposition 5.2.11, T® is the direct product of finite number of copies of C, and hence torsion-free Thus Tỷ is also torsion-free Note that for a split or ramified p the elements of 3„ differ from the elements of 3, defined in Proposition 5.2.7 only by multiplicative constants (but it is these constants that will make the action of Tỳ on the space of hermitian modular forms preserve the integrality of Fourier coefficients) In particular, one has T2, = T" Hence Tỷ is a finite free Z-algebra.
Lemma 5.2.13 Let £ be a rational prime, E a finite extension of Q¿ and O the valuation ring ring of E Suppose that F(Z) = })scr(B)e?ntr(P?) c S,(Tz) with cr(B) € O for all B Let T € TS Then TF(Z) = })pcrr(B)e?H(52) with crr(B) € O for every B.
Proof This follows directly from Lemma 5.2.4, Proposition 5.2.11 and Example 5.2.10 oO
Let AN denote a basis of eigenforms of S; (Tz).
Theorem 5.2.14 Let F € NTM There exists a number field Lp with ring of integers O,, such that Arc(T) € O1, for all T € Tô, Here Àrc(T) ts the eigenvalue of
Proof This is similar to the Eichler-Shimura isomorphism in the case of elliptic modular forms LÌ
Let £ be a rational prime and E a finite extension of Q¿ containing the fields Dr from Theorem 5.2.14 for all F € N* Denote by Ó the valuation ring of E and by À a uniformizer of O.
Definition 5.2.15 Let F € N° Then F gives rise to an O-algebra homomorphism
TỀ, — O assigning to T the eigenvalue of T corresponding to the eigenform F We denote this homomorphism by Ap Furthermore, we denote by Àz the composition of Ar with the natural projection O + F := Ó/À, and set mp := ker Àz.
Theorem 5.2.14 implies that we have
The O-algebra T}, also decomposes as a product of local rings This is completely analogous to the ellitpic modular case discussed in section 5.1 In particular the ideals mp (for F €.N*) exhaust the set of maximal ideals of T3, We have
To = [[ 73.0: hw h m where the product runs over the maximal ideals of TZ and T3 „ denotes the local- ization of TỀ, at m A similar description holds for TR, This proves the following corollary.
Corollary 5.2.16 If F € N® then there exists T € 'TÈ such that TF = F andTF’ = 0 for all eigenforms F” such that mp 2# mp, 0€, with not all eigenvalues congruent to those of F modulo À.
Action on the Maassspace 0000 cee eee eae 80
Theorem 5.2.17 (Gritsenko, [Gri90a], section 2) The action of the Hecke alge- bra T® respects the decomposition of S;,(Uz) into the Maass space and its orthogonal complement.
Theorem 5.2.18 (Gritsenko, [Gri90a], section 3) There exists a C-algebra map
Desc : Th > Tt such that for every T € T® the diagram
SMeass (Dz) —+— S55 (Tz) all [mrs Sear (4, (9) 45,1 (4, (=4)) commutes In particular one has
Desc(7.,) = p* “(1+ ứ?)T;z + p”~Š(p° + p?+p— 1) if pis inert in K, (5.14) Desc(T) = p*-?a-*(1+p)T, ifp= z7 is split in K,
Here T,, denotes the elliptic Hecke operator (5.1), and TY Tạ denotes the operator from Definition 5.1.9.
Corollary 5.2.19 If f € Š;_¡ (4, (=*)) is an eigenform, then so is Fy.
Proof This is immediate from Theorem 5.2.18 L]
Lifting Hecke operators to the Maass space 81 5.38 Adelic Hecke operators 2 ee va ga ga kg va 84 5.38.1 Localtheory HH gà ee 84
Let E and O be as before We will now prove a result regarding the map Desc, which will be used in section 7.4 Let Tz and T7 be as in Definition 5.1.9 For a discussion of the relevant properties of TZ, refer to section 8.1 It is clear from
Theorem 5.2.18 and the definition of Tz” that Desc(T')*) = T’, for any O-algebra
A Moreover, we have the following diagram
[n+ Tone — Tw Tom with the lower horizontal arrow defined so that the diagram commutes Let NV, as before, denote the basis of normalized eigenforms of S,_1 (4, (=*)) Recall that for f ¢N, f & ƒ? if and only if Fy # 0 The set VTM := {F, | g € N} is a basis of eigenforms of S)#5([z) by Corollary 5.2.19 Here, the superscript ‘M’ stands for ‘Maass’ Fix ƒ € MN with ƒ # f? Note that for g € N, m¿ = m¿ if and only if mp, = mỹ Here m; C To is the kernel of À¿|r„ (cf Definition 5.1.6), and mp, is the kernel of Àrlrs+ (cf Definition 5.2.15) The ideals m, and mp, are defined similarly All four are maximal ideals Hence Desc respects the direct product decomposition in diagram (5.15) In particular, Desc : T+ -ằ Tớ, factors through Tot, > Tomi, Let T3* be the image of TỆ” in Endoc(6}22%(Tz)).The horizontal arrows in diagram (5.15) factor through TB & ]]a Tàn and the following diagram
| | ⁄ | Met Toit —— la TO — Tw To, commutes All the horizontal arrows in diagram (5.16) are surjections and the lower ones are induced from the upper ones, which respect the direct product decomposi- tions The maximal ideals m7, m and m’ are defined in an obvious way In particular we have
Let NM! := {F € NTM | up, = tir} The goal of this section is to prove the following proposition.
Proposition 5.2.20 If fe N, ƒ # f? is ordinary at £, and £ > k, then for every split prime p = 17, pt £, there exists Th(p) € Toa such that Desc(T*(p)) € Tí, mở , Ff 3 equals the image of Tp € 'Tọẹ under the canonical projection To —> Tomi,’
As remarked before to every eigenform #` € Šx(ẽz) one can attach a 4-dimensional é-adic Galois representation pp As will be explained in section 10.1, if F = Fy, for some g € N, then the Galois representation has a special form
(Dg @ €) lơ where pg is the Galois representation attached to g (cf Theorem 10.1.2) and e is the f-adic cyclotomic character Let ƒ be as in Proposition 5.2.20 Set R’:= | [pc NM O and let R be the O-subalgebra of R’ generated by the tuples (Ar(T)) re AM for all
Te TS Here Àz : Tt — © is the O-algebra homomorphism sending T to the eigenvalue of 7' corresponding to the eigenform #' Then R is a complete Noetherian
83 local O-algebra with residue field F = O/X It is a standard argument to show that
Proof of Proposition 5.2.20 As remarked before, for ƒ,g € N, mà, = Mp, if and only if m, = mj, hence if mp, = mp, then m; = mj This implies that since f is ordinary at ý, so are all g € N such that Fy € NV 7! Let I, denote the inertia group at É Since g is ordinary at £, we have
(cf Theorem 10.1.2), and hence ek-2 x ĐF;ÌL, = 1 eel x
IfÊ > k it is easy to see that there exists ơ € J, such that the elements đị := e°~?(ứ),
By := 1, 83 := (0), By := €(c) are all distinct mod ý For every g as above, we choose a basis of the space of p, so that x = 0 Set
We extend p to an R-algebra map p’ : R[G| — M,(R’) Note that ỉứ(Erobze) = I] ỉr,(Erobz)/ứp, ( = [| 0g(Erobx)
FgeNM FyeNM and tr p'(Frob, e) = (a9(P)) ewe củ, where g = È}°.au(n)qg" Define 7°(p) to be the image of tr p/(Frob, e) under the O-algebra isomorphism R > Tore, O
Corollary 5.2.21 If ƒ € N, f # fe is ordinary at È, and Ê > k, then for every split prime p = TT, p†£, there exists Th(p) € TS + such that Desc(T*(p)) € Tạ mi
Mp ’ equals the image of Tp € To under the canonical projection To —> Tow,’
Proof We can choose TẦ(p) € Te to be any preimage of T*(p) € Tổ, from
Proposition 5.2.20 under the canonical Ó-algebra map Te — Tet, Oj
In this section we study the action of Hecke operators on the space of automor- phic forms and relate them to the classical operators defined so far We begin by considering the local situation The results of this and the following section are not needed in the sequel.
Let p be a finite prime of Q, and (V,7) an irreducible admissible representation of G(Q,) For a summary of the theory of such representations, see e.g [Bum97], section 4.2 Recall that Ko, := G(Z,) is the maximal (open) compact subgroup of G(Q,) Fix a left-invariant Haar measure dg on G(Q,) such that ƒ Kop 49 = 1 We denote by Hx,,, the algebra of compactly supported, smooth (i.e., locally constant), bi-Ko,-invariant functions from G(Q,) into C The multiplication in Hy,,, is defined to be the convolution
For any subset H of G(Q,) denote by [H} the characteristic function of H The function [Kp] is the identity element of Hx, ,.
Definition 5.3.1 The algebra Hx,,, is called the spherical Hecke algebra of G(Q,).
Lemma 5.3.2 The algebra Hx, 1s commutative.
Proof The proof is the same as the proof of Theorem 4.6.1 in [Bum97] O
The representation 7 defines an action v +> 7(¢)u of Hx,,, on V by m(d)u =| Ó(9)(g}0 dg.
G(Qp) Lemma 5.3.3 Let a € G(Q,) and assume that the double coset space Ko paKo,p decomposes as Ky paKoy = []j_, Kop with a; € G(Q,) Then for 0u €V we have x([KopaKog])u = À ` x(a)u ¡=1
Proof We have z(Ko,eKoy)s= | Ko,pakKo,p stovdg=S>f g)u dg. iKo,p
Making a change of variable g > aj‘g and using the fact that dg is a left-invariant measure we get
T([Kopao„] "NI (aig yoy = Yo aj)v
For any subgroup H of G(Q,), we denote by V” the subspace of V consisting of vectors fixed by H The elements of V*°ằ are called spherical vectors.
Lemma 5.3.4 The space V9? is at most one-dimensional.
Proof Since (V,7) is irreducible, it follows from Proposition 4.2.3 in [Bum97] that
VỀ is a simple Hx,,,-module Since Hx,, is commutative by Lemma 5.3.2, the claim follows L
Lemma 5.3.5 Letd€ Hx, and o € Vr Then r(ề})u € V9ằ,
Proof We need to show that 7(k)x(ở)0 = 1(ó)0 for k€ Kop We have x(k)r(ð} = m(k) [ ơ"
By Lemma 5.3.3 the integral over G(Q,) is just a finite sum, so we can interchange m(k) and Jứ(q,: Then x(R)r()ứ= | d(g)r(kg)odp
Making a change of variable g +> k”g and using the fact that dq is left-invariant we see that the last integral equals Je(a,) ó(k~1g)(g)u dg Since ¢ is left Ko „-invariant, the lemma follows L]
Lemma 5.3.6 Let a € G(Q,) If KopaKop = Lj, Kopai with a; € G(Qp), and v € V9z, then 3 m(a;`)u € V0,
Proof Note that if KopaKopy = []j_, Kopai, then Kopa Xeằ = |lÊ_Ăd;'Kop.
Hence by Lemma 5.3.3 demas" e = t([KopaTM’ Kopl)v. ¡=1
The claim now follows by Lemma 5.3.5 C]
In this section v will denote a (finite or infinite) place of Q Let g = gaqguk €G(Q)G(R)Ko¿(N) with Kor(N) = ]l¿ Kow(N) as before Every F € M3?(N, v)
87 gives rise to a function yr : G(A) — C defined as
@r(9) = (FlmGoo) (i) * (det dx), where Z = gui € H Let p # oo be a rational prime For a € G(Q) let ay) denote the element of G(A) whose p-component is a and all the other components are 1. Moreover, we will say that a is p-significant if the image of a under the natural embedding G(Q) — G(Q,) lands in Ko,,(.N) for all v # p, œ.
Lemma 5.3.7 Let p be a (finite) rational prime with p{ N Let F € MTM (N,v) and suppose a € G(Q) is p-significant Let g = (go, 1) € G(R) x G(At) be such that
(F ma) (Z) = 5 (Goo, i)" vn (det da)or (gaz):
(Flma) (2) = 5 (4s Goo)" F (aGooi) = 7(ags, i)" F (Goo: i)” F (Agee: 1) an = J (Goo i)TM Yr ((4g9o0s 1)) = F (Goo, i)” PF ((Goo a7", a7", -))
= j(goos i)" Yr ((Goo.1,1, -)(1,1, ,1,a4,1, )x x (lra7t,a7t, ,a7", 1,a7, )), where in the last expression a~! in the second factor and 1 in the third factor appear on the p-th place Note that (1,a7!,a71, ,a7+,1,a71, ) € Ko(N) as a is p- significant Thus we have
(F|ma) (Z) = 5 (Goo, i) "Yr (gag) Vn (det da).
Let a € G(Q) be p-significant Then [o(W)aTo(N) = []j_, Po(N)ai with a; €G(Q), p-significant In section 5.2.1 we defined an action of the double coset on MfPh(N,) Lemma 5.3.7 provides a tool to translate this action to an action on functions of the form yr as defined above.
Corollary 5.3.8 Let p be a (finite) rational prime with pt N Let F € MUỤR(N,0) and let a € G(Q) be p-significant Let g = (gx.1) € G(R) x G(Ai) be such that
(Flm[fo(N)aTo(W)]) (Z) = 9 (Goo, 1) "Dvr ga;,))-
Proof The corollary follows immediately from Lemma 5.3.7 and the definition of the action of Fe(W)aTo(N) on MIP(N,ð) Oo
5.4 Hecke Eigenvalues of Siegel Eisenstein series
In this section we compute the eigenvalue of the operator T® defined by (5.6) on the Siegel Eisenstein series of positive weight, level and non-trivial character
(cf section 3.2.4) The eigenvalues of TP, for p inert, on Siegel Eisenstein series of level 1 were computed in [Kri91] Even though we will not make use of these calculations later on, they would be useful if one would like to generalize the theory of Maass lifts to hermitian modular forms of level higher than one We will use notation from section 3.2.4 In this section v will denote a place of Q First note that E(Z,s,m,v,N) € M2®(N, 71), hence it makes sense to talk about the action of T? on E(Z,s,m, wv, N).
Proposition 5.4.1 Let p= a7 be a prime of Q which splits in K and assume that ĐỊN Then the series E(Z,s,m,t,N) defined by formula (3.17) is an eigenform for T® with eigenvalue p?m~$-m[3—m (0 4 1)2x(a) 1 (by (p)*p?$ +7°)
1 Hence by (5.8) we obtain a right coset decomposition of 7o of the form
We now study the action of T?') on E(Z, s,m,, N) Let g = (goo, 1) € G(R)xG(A;) be such that Z = gyi Then E(g,s,N,m,v) = j(g0,i1) "E(Z, s,m,v, N) Note that if o1 ẤN, then Ko,(N) is the maximal compact subgroup of G(Q,) We also have (cf section 3.2.4)
E(g,s,N,mv)= So) tp(g)ụp(ag)°)2é= So dz+ 39)ụps(xứ,)°” yEP(Q)\G(Q) yEP(Q)\G(Q) ¥ with pp, and dp, as in (3.15) and (3.16) For g’ € G(A) set f(g’, 8) = Upo(9')OPa(g')”.
Denote by A the set of right coset representatives from (5.19) Note that allae A are p-significant, hence by Corollary 5.3.8 we have
Thus we are reduced to studying the right-hand side of (5.20) For g € G(A) we have ằ 90, ,8, í,m, Ú) = S]]z (2e(2(ứ))x ` 8) = acA acA v
~ =>, 1I2‹( 9v:8) | Íứ(0p8”`,8 = |]H] 0.5) À ` /p(gpa~",8) acA \ up utp acA
Note that we have Ko(N) = Koy, the maximal compact subgroup of G(Q,), and (by the same method as (5.8) was obtained) we have
Kop 1 z 1 Kop = I] Ko pa na} acA
Since f,(-,s) is right Ko,-invariant, Lemma 5.3.6 implies that >,.4 fp(gpa7", 8) is also right Ko,-invariant Thus by Lemma 5.3.4 there exists À € C such that À ` folGpa*, 8) = Afp(9p; 8). acA
Since A is independent of gp, we have A = ƒp;(1,s) '5 ).
Note that Np = 2 Now, taking the product over all the finite places we see that we obtain the Euler factors for the right-hand side of (6.3) L
So far we have studied the numerator (Fy, E(-,8,m, T*)6,) (Chapter 6), and the denominator (Fy, F;) (Chapter 4) of the coefficient Cr, in formula (1.2) We ex- pressed them by L-functions associated to ƒ To use these formulas for studying congruences between Fy and other hermitian modular forms, we need to first investi- gate the integrality of Fourier coefficients of E(-,s,m,I"") and of 6, This is done in sections 7.1 and 7.2 We state the first main theorem of this thesis (Theorem 7.4.5) in section 7.4.
7.1 Fourier coefficients of Eisenstein series
We keep the notation from section 6.1 and assume b = 1 Consider the set X of Hecke characters x’ of K, such that gm
(7.2) X;(Zp) = 1 if p Joo, tp € OF, and zp — 1 € cOxy.
Here m= k—1=-t—2> 0 (since ý < —6) denotes the weight of the Eisenstein series E(Z,s,m,I") defined in section 6.1 For g € G(A), let E(g, s,¢,m, x’) denote the Siegel Eisenstein series defined in section 3.2.4 We put, as before,
101 where Z = gyi and g = (gs,l) Recall from section 6.1 that we made use of a congruence subgroup I? of G(Q) such that 0, € A,(F}) and F}nK* = {1} In this section we fix a particular choice of ẽ*, namely, we set [> := I}(c) Note that as long as c† 2, we have [?(c) M K* = {1} and since (cond wv”) | c, where wv’ is the character of 6y, we have 6, € ,M1,(F}(c)) Lemma 17.2 in [Shi00] provides a connection between E(Z,s,m,x',e) and E(Z,s,m,T2(c)) Because the lemma in [Shi00] is stated in a context slightly different from ours, we quote it here in a form which is applicable to our situation.
Lemma 7.1.1 Let X be the set of all Hecke characters of K satisfying (7.1) and
= ằ ằ E(Z2,s,m,x,€) |m @. x'€X aeTh(e)\F} (e) Note that the map
T2) \TỆ() = Ma(Ox/cỉx) given by y+ b„ (mod cOx) is a bijection In fact we can choose a set of representa- tives for I*(e) \T2(c) in the form [7%] with b, € M2(Ox) and thus get
(7.4) ằ E(Z,s,m,x',Â) |m œ =| Ox/cOx |* E(Z,s,m,x',c) = CE(Z,s,m,x’,c), aéTh (e)\TB(e) since E(Z,s,m,x',c) |m a = E(Z, s,m, x’, c)x’(det d.) for all a € To(c).
Definition 7.1.2 Let M be a non-zero integer For a Hecke character / : Q*\A* > C”* we set
Lu(s,¥) = L(s,9) ][ — 0 pn, p|M where E(s, ý) denotes the Dirichlet £-function.
Recall that for any Hecke character ý : #Z* \ AX => C* we denote by vq its restriction to A*.
It has been shown in [Shi00] (Theorem 17.12(11)) that D(Z, s,m, x’,c) is holomorphic in the variable Z for s = 2— 5 as long as m > 2 In our case m = —t —2 > 4 as t < —6 The transformation property
(7.5) E(agw, s,c,m, x’) = x¿(det dy)71j(w, i) 1 E(g, 8,¢,m, x’), where a € G(Q) and w € Ko(c), proved in [Shi97] (formula (18.6.2)) ensures that
Instead of looking at D(Z, s,m, x’, c) we will study the Fourier expansion of a trans- form D*(Z, s,m, x’,c) defined by
(7.7) D*(Z,8,m,x',c) = D(Z, s,m, x’, C6) |md, where J = E “| First note that since D is holomorphic at s = 2 — 4, so is ~1
D*(Z,2 —m/2,m,x’,c) = Sok (2 — =) e(trhZ) hes 2 for the Fourier expansion of D* More generally we can write
D'(Z,s,m,x,e) = ằ cX (Y,s)e(trhX), hes where X = Re Z and Y = Im Z If for a given s, D* is holomorphic in Z, then
One can also write a “Fourier expansion” for the adelic transform of
D(g, 8,¢,m, x’) = L(28, XQ) Lc (2s —1Xx@ (=)) E(g,8,¢,m, x’) defined by
D*(g,8,c,m,x’) = D(gJy",¢,8,m, x’), where J, is the matrix in G(A) which is 1 at the infinite place and equals J at all the finite places The Fourier expansion takes the form:
D* ,8,€,Tn, X | = À `c(h, qg, s)ea(tr (hứ)),
(ứ) 1 hes where ơ € S(A) := {œ € 4⁄;(A) | (@)' = a} and g € GL2(K) with detg = 1 For the definition of the exponential eA see [Shi00], section 1.6 We will not need it here.
(7.9) en(Y,s) = (det ¥)7"2c(h, Y/?, 8), where Y1⁄2 is a matrix in S(A) whose oo-component is the positive-definite matrix whose square is Y and all the finite components are 1 So, we are reduced to studying c(h,Y1⁄2, s).
1—rank(h) —4 j-l x J] L (2 —m— jx’ (=) II j=0 pec where fi, y1/2,, 1s 6 polynomial with coefficients in Z and constant term 1, and c is a certain finite set of primes Ifn WO’) ) (det ay) =
+€T?(e)\Tÿ(e) 0 otherwise, we conclude that (7.16) equals
AX SO BZ, s,m, ve) T5(c) THO] = £[T(e) : T1(e)]E(2, s,m, vc) = x'.ex
(E(., Đ,TM, T2(c)) 4, FF) pe (6) — CPS (c) , T?(e)] (EC, 8; mv’, c)ỉx, FF) peo) ,
(7.18) (DG,s,m,v’, €)Ox; Ft) re =c “TẠ() : Te(o)]7 (2s, Hq) x
Moreover, by |Shi00], formula (17.5) and Remark 17.12(2), we have
(7.19) ŒC, s,m, TY (€)) Ox; Fp) mu =[f():P! (E(-,8,m, D2 (c)) Ox: Fp) rn =
= [T}(e) : TP? l ằ E(Z, s,m, an) OF) œ€T\TẺ (c) rh
Since 6,|;@ = 6, and Ƒ;|¿œ = Fy for œ € I}(c) we finally have
TROT 2, (om Mina) li) Ahan =
(7.21) (D(:,s,m,,c)9x, Fs) pe (6) = (c'AB(2 - m/2)) © x L¿(2s, ỦQ) Le (2s —1,vQ (=)) (det r)~* x cr, (7) Le (Fy, 3 + 1, x)
= c~*B(2 — m/2)~}(det 7)” (4n)? T (m') F (mĩ — 1) x cr, (T)La(F,3—m/2,x) lim L(s),
We have (úq)"” = xq Hence ¥q = (xq)”” = ỦQ as all Hecke characters are assumed to be unitary Thus L(s) = 1 Moreover
(723) (D(,2~ mỊ2,m, Wc), Ft)p(¿ = Ra ? 3D + k + 2) +k + 1)x x Ly (F, 3 — m/2, x),where R := er,(7)c"*B(2 — m/2)~} (8 detr)~?~*~2.
In this section we prove the first main result of this thesis We will show that é-divisibility of L(Symm? ƒ, k) implies the existence of a non-Maass cusp form con- gruent to Fy modulo 2 We keep the notation from previous sections.
Fourier coefficients of Eisenstein SeTÌS HQ ee 101
We keep the notation from section 6.1 and assume b = 1 Consider the set X of Hecke characters x’ of K, such that gm
(7.2) X;(Zp) = 1 if p Joo, tp € OF, and zp — 1 € cOxy.
Here m= k—1=-t—2> 0 (since ý < —6) denotes the weight of the Eisenstein series E(Z,s,m,I") defined in section 6.1 For g € G(A), let E(g, s,¢,m, x’) denote the Siegel Eisenstein series defined in section 3.2.4 We put, as before,
101 where Z = gyi and g = (gs,l) Recall from section 6.1 that we made use of a congruence subgroup I? of G(Q) such that 0, € A,(F}) and F}nK* = {1} In this section we fix a particular choice of ẽ*, namely, we set [> := I}(c) Note that as long as c† 2, we have [?(c) M K* = {1} and since (cond wv”) | c, where wv’ is the character of 6y, we have 6, € ,M1,(F}(c)) Lemma 17.2 in [Shi00] provides a connection between E(Z,s,m,x',e) and E(Z,s,m,T2(c)) Because the lemma in [Shi00] is stated in a context slightly different from ours, we quote it here in a form which is applicable to our situation.
Lemma 7.1.1 Let X be the set of all Hecke characters of K satisfying (7.1) and
= ằ ằ E(Z2,s,m,x,€) |m @. x'€X aeTh(e)\F} (e) Note that the map
T2) \TỆ() = Ma(Ox/cỉx) given by y+ b„ (mod cOx) is a bijection In fact we can choose a set of representa- tives for I*(e) \T2(c) in the form [7%] with b, € M2(Ox) and thus get
(7.4) ằ E(Z,s,m,x',Â) |m œ =| Ox/cOx |* E(Z,s,m,x',c) = CE(Z,s,m,x’,c), aéTh (e)\TB(e) since E(Z,s,m,x',c) |m a = E(Z, s,m, x’, c)x’(det d.) for all a € To(c).
Definition 7.1.2 Let M be a non-zero integer For a Hecke character / : Q*\A* > C”* we set
Lu(s,¥) = L(s,9) ][ — 0 pn, p|M where E(s, ý) denotes the Dirichlet £-function.
Recall that for any Hecke character ý : #Z* \ AX => C* we denote by vq its restriction to A*.
It has been shown in [Shi00] (Theorem 17.12(11)) that D(Z, s,m, x’,c) is holomorphic in the variable Z for s = 2— 5 as long as m > 2 In our case m = —t —2 > 4 as t < —6 The transformation property
(7.5) E(agw, s,c,m, x’) = x¿(det dy)71j(w, i) 1 E(g, 8,¢,m, x’), where a € G(Q) and w € Ko(c), proved in [Shi97] (formula (18.6.2)) ensures that
Instead of looking at D(Z, s,m, x’, c) we will study the Fourier expansion of a trans- form D*(Z, s,m, x’,c) defined by
(7.7) D*(Z,8,m,x',c) = D(Z, s,m, x’, C6) |md, where J = E “| First note that since D is holomorphic at s = 2 — 4, so is ~1
D*(Z,2 —m/2,m,x’,c) = Sok (2 — =) e(trhZ) hes 2 for the Fourier expansion of D* More generally we can write
D'(Z,s,m,x,e) = ằ cX (Y,s)e(trhX), hes where X = Re Z and Y = Im Z If for a given s, D* is holomorphic in Z, then
One can also write a “Fourier expansion” for the adelic transform of
D(g, 8,¢,m, x’) = L(28, XQ) Lc (2s —1Xx@ (=)) E(g,8,¢,m, x’) defined by
D*(g,8,c,m,x’) = D(gJy",¢,8,m, x’), where J, is the matrix in G(A) which is 1 at the infinite place and equals J at all the finite places The Fourier expansion takes the form:
D* ,8,€,Tn, X | = À `c(h, qg, s)ea(tr (hứ)),
(ứ) 1 hes where ơ € S(A) := {œ € 4⁄;(A) | (@)' = a} and g € GL2(K) with detg = 1 For the definition of the exponential eA see [Shi00], section 1.6 We will not need it here.
(7.9) en(Y,s) = (det ¥)7"2c(h, Y/?, 8), where Y1⁄2 is a matrix in S(A) whose oo-component is the positive-definite matrix whose square is Y and all the finite components are 1 So, we are reduced to studying c(h,Y1⁄2, s).
1—rank(h) —4 j-l x J] L (2 —m— jx’ (=) II j=0 pec where fi, y1/2,, 1s 6 polynomial with coefficients in Z and constant term 1, and c is a certain finite set of primes Ifn WO’) ) (det ay) =
+€T?(e)\Tÿ(e) 0 otherwise, we conclude that (7.16) equals
AX SO BZ, s,m, ve) T5(c) THO] = £[T(e) : T1(e)]E(2, s,m, vc) = x'.ex
(E(., Đ,TM, T2(c)) 4, FF) pe (6) — CPS (c) , T?(e)] (EC, 8; mv’, c)ỉx, FF) peo) ,
(7.18) (DG,s,m,v’, €)Ox; Ft) re =c “TẠ() : Te(o)]7 (2s, Hq) x
Moreover, by |Shi00], formula (17.5) and Remark 17.12(2), we have
(7.19) ŒC, s,m, TY (€)) Ox; Fp) mu =[f():P! (E(-,8,m, D2 (c)) Ox: Fp) rn =
= [T}(e) : TP? l ằ E(Z, s,m, an) OF) œ€T\TẺ (c) rh
Since 6,|;@ = 6, and Ƒ;|¿œ = Fy for œ € I}(c) we finally have
TROT 2, (om Mina) li) Ahan =
(7.21) (D(:,s,m,,c)9x, Fs) pe (6) = (c'AB(2 - m/2)) © x L¿(2s, ỦQ) Le (2s —1,vQ (=)) (det r)~* x cr, (7) Le (Fy, 3 + 1, x)
= c~*B(2 — m/2)~}(det 7)” (4n)? T (m') F (mĩ — 1) x cr, (T)La(F,3—m/2,x) lim L(s),
We have (úq)"” = xq Hence ¥q = (xq)”” = ỦQ as all Hecke characters are assumed to be unitary Thus L(s) = 1 Moreover
(723) (D(,2~ mỊ2,m, Wc), Ft)p(¿ = Ra ? 3D + k + 2) +k + 1)x x Ly (F, 3 — m/2, x),where R := er,(7)c"*B(2 — m/2)~} (8 detr)~?~*~2.
In this section we prove the first main result of this thesis We will show that é-divisibility of L(Symm? ƒ, k) implies the existence of a non-Maass cusp form con- gruent to Fy modulo 2 We keep the notation from previous sections.
Let ý { 2c be a rational prime, and let E and Ó be as in Corollary 7.2.2 Fix a uniformizer À € O We denote the A-adic valuation by ord) Several times in this section we will replace by a finite extension and denote that extension again by
E Every time we do so, we assume that © and ord) are also redefined in an obvious way In particular we always normalize ord, in such a way that ord)(A) = 1 Recall formula (7.12):
(D(.,2 — m/2,m, ', €) Oy, Fy) pn đụ, Fs) px (6) for F’ € MI,(c) with (F;, F’) = 0 We can clearly drop the subscripts LỆ(c) on the
DỰ, 2~ m/2,.m, ÿ', e) (Z) inner products as long as we consider their quotients We have
+ F'|pJ, where J = B 2 ] and I, denotes the 2 x 2-identity matrix Since J € Tz, and F; is of full level, we have Fy|,J = F; Using the notation from (7.7), we have
From now on we fix m, w’, e and Z and put
D* = D*(Z,2 —m/2,m,v’,0), and Œ' = F”|¿J Moreover as D6, € Mz(T$(c)), we have D*6,|¡J € Mz (J7'T§(c) J).
(Fy, G") japr(ay = (Fyled, GJ) japn(ea = (Fy, Œrs( = 0.
D*(0J) = tên A Fe + G!’ (Fy, Fy) ; with (F;,G’) =0 By Corollary 7.2.2, the Fourier coefficients of
3 Atth+2)+9 (det, T)?D*(6,[J) all lie in O Thus so do the Fourier coefficients of 7~3(det 7)?.D*(6,|)J), and the Fourier coefficients of both sides of
Dé,, Fs) nm *(det T)’D*(Oy\1J) = ` z-”(det7)?F + x7”(det)?ŒI. địt)
Define a trace operator tr: My,( JT 2 (c) J) — M,(Tz) via tũ ằ Fy+.
7.25) = := tra73(det r)?D*(6,|)J) = tra? (det 7 2 (DO, Fs) F;+tr 7~3(det 7)°G’ * (Fy, Fy) 7
Both sides are modular forms of full level Moreover, as F¥ itself is of full level we get: tr Fy = (Pz: J-!1TẠ(e)J] Fy = (Pz : Tậ(e)] Fy.
Set G” = trz~”(det7)?Œ! € M,(T'z) and note that Œ,G) = SO (FG ey) = CS (Fela, @ ley) = ¢ SFG) = 0,
By the g-expansion principle (Theorem 2.3.2), the Fourier coefficients of = and hence also of
[Pz : Pg (e)|a 7° (det 7) ‘UF; Fi) tr+G lie in Ó Set ơ e\q-3 2(DOx, Fs)
Cr, := [fz : Tộ(e)]x"”(det 7) (F;, Fy)
Remark 7.4.1 If gcd({a¡;}) = 1, where 7 = , then ord, (det 7) > 0 only
421 22 if p | c, so we can also drop the (det r)-factor.
By Proposition 4.1.5, the Fourier coefficients of #; are algebraic integers and generate a finite extension of Q We enlarge EF, so it contains all the Fourier coefficients of
(D0x,F;) ri) respectively Let B(s) be the function defined in section 6.1.
F; The numerator and denominator of were studied in sections 7.3 and 4.2
(Dx: E2) (4) „ DETR + 2, Fgax)DUE + k +1, F ga)
Vàng), L*'8(Symm? ƒ, k) ị where œ:= B(2— m/2)7* (det 7) cp, (7),
111813 (7) L(Symm? ƒ,n) mnt? (f, f) for any integer n, and (*) € QNE is a d-adic unit Here D(s, ƒ, gay) is as in Remark
Proof This is a straightforward calculation using formulas (4.2.4) and (7.23) LÌ
It follows from Theorem 1 on page 325 in [Hid93] that
(7.28) D*S(+k+2, ƒ, gaz) Q and from a result of Sturm [Stu80] that
We note here that [Stu80] uses a definition of the Petersson norm of ƒ which differs from ours by a factor of `, the volume of the fundarnental domain for the action of SL2(Z) on the complex upper half-plane We now enlarge the field # to include values (7.27), (7.28), and (7.29).
As we are ultimately interested in (mod A) congruences between hermitian mod- ular forms, we need to use “integral periods” OF, Q; instead of (f, f) These are defined in section 8.3 and it follows from Proposition 8.3.4 that we have:
(ff) = (#) nQFQ5, where ? is an algebraic integer called the Hida invariant (cf Definition 8.3.2) and (x) is a A-adic unit We enlarge #2 again so it contains 7.
Proof This follows directly from Lemma 7.4.2 upon noting that orda(B(2— m/2)) >
7.4.3 Congruence between F; and a non-Maass form
As our main goal is to prove that Fy is congruent to a non-Maass form, we need to find a way to kill the “Maass part” of G” in formula
We do so by using the properties of appropriate Hecke algebras established in chapters
Let Tz and T7 be as in Definition 5.1.9 Let MN be the basis of normalized eigenforms in Sp_1 (4, (=*)) For g EN, T € Tz, let À;¿c(7) be the eigenvalue of 7 corresponding to g By Theorem 5.1.4 the set
(7.31) Y:= {A\,.c(T) |g Ee N,T € Tz} is contained in the ring of integers of a finite extension of Q.
On the other hand, by Proposition 5.2.11, the space SŠ;(Tz) also has a basis A/* consisting of eigenforms Let TS, Ty be as in Definition 5.2.12 By Theorem 5.2.14, the set
(7.32) DO := {\pc(T)| FE N*,T € TH} is contained in the ring of integers of a finite extension of Q Here Àrc(7) denotes the eigenvalue of 7' corresponding to F We enlarge E, so it contains the set UUD®. From now on assume that f is ordinary at ý, and that ứ;|ứ„ is absolutely irreducible. Here ứ; denotes the (mod Â) Galois representation attached to f (cf section 10.1). For any F € N°, let mp be as in Definition 5.2.15 By Corollary 5.2.16 there exists T* € Tộ such that ThƑ; = Fy and T"F = 0 for all F € NTM such that mp # mự,.
We apply 7? to both sides of
As Fourier coefficients of Fy and & lie in O, so do the Fourier coefficients of T"= by Lemma 5.2.13 Moreover, since 6, is a cusp form, so are = and T= Let Sy rc S;,(Uz) denote the subspace spanned by
Here mp and mỹ, denote the maximal ideals of T3* defined in section 5.2.5 (see the discussion following diagram (5.15)) Then 7°E,T*Ƒ; = Fy,T"G” € Sy , The image of TS" inside Endo (Sy ;) can be naturally identified with Tom, By the commutativity of diagram (5.15) and the discussion following the diagram, the O- algebra map Desc : TS" ~ằ To factors through Tomi, > Tomi The algebra
Tom, can be identified with the image of T inside Endc(S_1,¢), where Sp—1,¢ CSz-1 (4, (=)) is the subspace spanned by NV} := {g € NV | m, = mj} Here mi and m¿ denote the maximal ideals of To defined in section 5.2.5 (see the discussion following diagram (5.15)) Denote by ó¿ the natural projection To => To and by â, the natural projection Ty" —ằ Tomộ, Assume Ê > k By Corollary 5.2.21, for every split prime p = 77, p { £, there exists T"(p) € TO mt, such that Desc(T"(p)) = bs (Tp) € Tom, As will be shown in chapter 8 (cf Proposition 8.3.3), there exists a Hecke operator T € Tom, such that Tf = nf, Tf? = nf’, and Tg = 0 for all gcM.,gz f, f°’ The operator T is a polynomial Pr in the elements of ó;(5) with coefficients in O Let T € Tot, be the Hecke operator given by the polynomial P; obtained from Pr by substituting e ỉ;(T — pk? — p-? — p°~3) for úg(T;2) if p inert in K, e Th(p) for ở/(7p) if p{ £ splits in K, e ®/(Aÿ2-*(+ 1) 1T? ) for @(1;) if £ = AoAo splits in K.
For the definition of operators 7" and Ty, see formulas (5.5) and (5.6) respectively, and for the definition of T"(p) see Proposition 5.2.20 Note that Z2 *(/+ 1) 17 is indeed an element of Ty mt 38 £+ 1 is invertible in O It follows from formulas
(5.14) that Desc(T) = T Apply T to both sides of
Note that T75 is again a cusp form The operator T preserves the Maass space and its orthogonal complement by Theorem 5.2.17 Another application of Lemma
5.2.13 shows that the Fourier coefficients of TT*S lie in O Moreover, since Desc is a C-algebra map, it is clear from the definition of T that TF ?= ni; and TFP=0 for any F' inside the Maass space of S,([I'z) which is orthogonal to Fy, i.e., such that(F, Fy) = 0 As Cr, € QNE CC by Corollary 7.4.3, it makes sense to talk about
Main congruence result 2 ee 112
In this section we prove the first main result of this thesis We will show that é-divisibility of L(Symm? ƒ, k) implies the existence of a non-Maass cusp form con- gruent to Fy modulo 2 We keep the notation from previous sections.
Let ý { 2c be a rational prime, and let E and Ó be as in Corollary 7.2.2 Fix a uniformizer À € O We denote the A-adic valuation by ord) Several times in this section we will replace by a finite extension and denote that extension again by
E Every time we do so, we assume that © and ord) are also redefined in an obvious way In particular we always normalize ord, in such a way that ord)(A) = 1 Recall formula (7.12):
(D(.,2 — m/2,m, ', €) Oy, Fy) pn đụ, Fs) px (6) for F’ € MI,(c) with (F;, F’) = 0 We can clearly drop the subscripts LỆ(c) on the
DỰ, 2~ m/2,.m, ÿ', e) (Z) inner products as long as we consider their quotients We have
+ F'|pJ, where J = B 2 ] and I, denotes the 2 x 2-identity matrix Since J € Tz, and F; is of full level, we have Fy|,J = F; Using the notation from (7.7), we have
From now on we fix m, w’, e and Z and put
D* = D*(Z,2 —m/2,m,v’,0), and Œ' = F”|¿J Moreover as D6, € Mz(T$(c)), we have D*6,|¡J € Mz (J7'T§(c) J).
(Fy, G") japr(ay = (Fyled, GJ) japn(ea = (Fy, Œrs( = 0.
D*(0J) = tên A Fe + G!’ (Fy, Fy) ; with (F;,G’) =0 By Corollary 7.2.2, the Fourier coefficients of
3 Atth+2)+9 (det, T)?D*(6,[J) all lie in O Thus so do the Fourier coefficients of 7~3(det 7)?.D*(6,|)J), and the Fourier coefficients of both sides of
Dé,, Fs) nm *(det T)’D*(Oy\1J) = ` z-”(det7)?F + x7”(det)?ŒI. địt)
Define a trace operator tr: My,( JT 2 (c) J) — M,(Tz) via tũ ằ Fy+.
7.25) = := tra73(det r)?D*(6,|)J) = tra? (det 7 2 (DO, Fs) F;+tr 7~3(det 7)°G’ * (Fy, Fy) 7
Both sides are modular forms of full level Moreover, as F¥ itself is of full level we get: tr Fy = (Pz: J-!1TẠ(e)J] Fy = (Pz : Tậ(e)] Fy.
Set G” = trz~”(det7)?Œ! € M,(T'z) and note that Œ,G) = SO (FG ey) = CS (Fela, @ ley) = ¢ SFG) = 0,
By the g-expansion principle (Theorem 2.3.2), the Fourier coefficients of = and hence also of
[Pz : Pg (e)|a 7° (det 7) ‘UF; Fi) tr+G lie in Ó Set ơ e\q-3 2(DOx, Fs)
Cr, := [fz : Tộ(e)]x"”(det 7) (F;, Fy)
Remark 7.4.1 If gcd({a¡;}) = 1, where 7 = , then ord, (det 7) > 0 only
421 22 if p | c, so we can also drop the (det r)-factor.
By Proposition 4.1.5, the Fourier coefficients of #; are algebraic integers and generate a finite extension of Q We enlarge EF, so it contains all the Fourier coefficients of
(D0x,F;) ri) respectively Let B(s) be the function defined in section 6.1.
F; The numerator and denominator of were studied in sections 7.3 and 4.2
(Dx: E2) (4) „ DETR + 2, Fgax)DUE + k +1, F ga)
Vàng), L*'8(Symm? ƒ, k) ị where œ:= B(2— m/2)7* (det 7) cp, (7),
111813 (7) L(Symm? ƒ,n) mnt? (f, f) for any integer n, and (*) € QNE is a d-adic unit Here D(s, ƒ, gay) is as in Remark
Proof This is a straightforward calculation using formulas (4.2.4) and (7.23) LÌ
It follows from Theorem 1 on page 325 in [Hid93] that
(7.28) D*S(+k+2, ƒ, gaz) Q and from a result of Sturm [Stu80] that
We note here that [Stu80] uses a definition of the Petersson norm of ƒ which differs from ours by a factor of `, the volume of the fundarnental domain for the action of SL2(Z) on the complex upper half-plane We now enlarge the field # to include values (7.27), (7.28), and (7.29).
As we are ultimately interested in (mod A) congruences between hermitian mod- ular forms, we need to use “integral periods” OF, Q; instead of (f, f) These are defined in section 8.3 and it follows from Proposition 8.3.4 that we have:
(ff) = (#) nQFQ5, where ? is an algebraic integer called the Hida invariant (cf Definition 8.3.2) and (x) is a A-adic unit We enlarge #2 again so it contains 7.
Proof This follows directly from Lemma 7.4.2 upon noting that orda(B(2— m/2)) >
7.4.3 Congruence between F; and a non-Maass form
As our main goal is to prove that Fy is congruent to a non-Maass form, we need to find a way to kill the “Maass part” of G” in formula
We do so by using the properties of appropriate Hecke algebras established in chapters
Let Tz and T7 be as in Definition 5.1.9 Let MN be the basis of normalized eigenforms in Sp_1 (4, (=*)) For g EN, T € Tz, let À;¿c(7) be the eigenvalue of 7 corresponding to g By Theorem 5.1.4 the set
(7.31) Y:= {A\,.c(T) |g Ee N,T € Tz} is contained in the ring of integers of a finite extension of Q.
On the other hand, by Proposition 5.2.11, the space SŠ;(Tz) also has a basis A/* consisting of eigenforms Let TS, Ty be as in Definition 5.2.12 By Theorem 5.2.14, the set
(7.32) DO := {\pc(T)| FE N*,T € TH} is contained in the ring of integers of a finite extension of Q Here Àrc(7) denotes the eigenvalue of 7' corresponding to F We enlarge E, so it contains the set UUD®. From now on assume that f is ordinary at ý, and that ứ;|ứ„ is absolutely irreducible. Here ứ; denotes the (mod Â) Galois representation attached to f (cf section 10.1). For any F € N°, let mp be as in Definition 5.2.15 By Corollary 5.2.16 there exists T* € Tộ such that ThƑ; = Fy and T"F = 0 for all F € NTM such that mp # mự,.
We apply 7? to both sides of
As Fourier coefficients of Fy and & lie in O, so do the Fourier coefficients of T"= by Lemma 5.2.13 Moreover, since 6, is a cusp form, so are = and T= Let Sy rc S;,(Uz) denote the subspace spanned by
Here mp and mỹ, denote the maximal ideals of T3* defined in section 5.2.5 (see the discussion following diagram (5.15)) Then 7°E,T*Ƒ; = Fy,T"G” € Sy , The image of TS" inside Endo (Sy ;) can be naturally identified with Tom, By the commutativity of diagram (5.15) and the discussion following the diagram, the O- algebra map Desc : TS" ~ằ To factors through Tomi, > Tomi The algebra
Tom, can be identified with the image of T inside Endc(S_1,¢), where Sp—1,¢ CSz-1 (4, (=)) is the subspace spanned by NV} := {g € NV | m, = mj} Here mi and m¿ denote the maximal ideals of To defined in section 5.2.5 (see the discussion following diagram (5.15)) Denote by ó¿ the natural projection To => To and by â, the natural projection Ty" —ằ Tomộ, Assume Ê > k By Corollary 5.2.21, for every split prime p = 77, p { £, there exists T"(p) € TO mt, such that Desc(T"(p)) = bs (Tp) € Tom, As will be shown in chapter 8 (cf Proposition 8.3.3), there exists a Hecke operator T € Tom, such that Tf = nf, Tf? = nf’, and Tg = 0 for all gcM.,gz f, f°’ The operator T is a polynomial Pr in the elements of ó;(5) with coefficients in O Let T € Tot, be the Hecke operator given by the polynomial P; obtained from Pr by substituting e ỉ;(T — pk? — p-? — p°~3) for úg(T;2) if p inert in K, e Th(p) for ở/(7p) if p{ £ splits in K, e ®/(Aÿ2-*(+ 1) 1T? ) for @(1;) if £ = AoAo splits in K.
For the definition of operators 7" and Ty, see formulas (5.5) and (5.6) respectively, and for the definition of T"(p) see Proposition 5.2.20 Note that Z2 *(/+ 1) 17 is indeed an element of Ty mt 38 £+ 1 is invertible in O It follows from formulas
(5.14) that Desc(T) = T Apply T to both sides of
Note that T75 is again a cusp form The operator T preserves the Maass space and its orthogonal complement by Theorem 5.2.17 Another application of Lemma
5.2.13 shows that the Fourier coefficients of TT*S lie in O Moreover, since Desc is a C-algebra map, it is clear from the definition of T that TF ?= ni; and TFP=0 for any F' inside the Maass space of S,([I'z) which is orthogonal to Fy, i.e., such that(F, Fy) = 0 As Cr, € QNE CC by Corollary 7.4.3, it makes sense to talk about
119 its \-adic valuation Suppose ord,(7 Cr,) = —n € Zọ Note that since the Fourier coefficients of TT*S and of F; lie in O, but nCr, ¢ O, we must have TT"G" £0 Write 7 Cr, = a\~” with a € O* Then the Fourier coefficients of MTTG" lie in
As explained above, —a~!\"T®TG” is a hermitian modular form orthogonal to the
We have proven the following theorem:
Theorem 7.4.5 Let S = {h € K3|h' =h} Let £ be a rational prime and k < £ a positive integer divisible by 4 Let f € Spi (4, (=)) be a normalized eigenform ordinary at £ such that Pr|o„ 1s absolutely irreducible where py denotes the (mod £) Galois representation attached to f Suppose there exists a2 x 2 matrir 7 € S such that ord¿(er,(7)) = 0, {ÿ79};eo; D Ox, and such that there exists c € Z with
(c’,£) = 1 and {(9°)*77"g}geo2, C (ở) 7O Furthermore, assume that there exists a Hecke character x : KX \ AX — C* with conductor prime to Ê and such that
Xo(z) = () 7 with —k < t < -6 Set c:= Œ Nx/q(condy) Let E denote a finite extension of Qe with uniformizer À, containing K(x¿, nạ) U 3S, the ele- ments (7.27), (7.28), (7.29), the Fourier coefficients of Fy and the Hida invariant n. Suppose that
—n := ord, (I DTM(t+k+5,f, on — ord,(L'TM(Symm? ƒ, k)) < 0 2 j=l where w is the unique Hecke character of K which 1s unramified at all fintte places and such that wo.(z) = ( š )” lz|
Then there exists a non-CAP cusp form F” such that ord,(Fy — F’) >n > 0.
Remark 7.4.6 By Proposition 8.2.10, the absolute irreducibility of ỉ;|œ„ implies that f # ƒ“ mod A, hence in particular #; # 0.
Remark 7.4.7 The conditions on the matrix 7 ensure that we can take b = 1, where b denotes the constant defined in section 6.1.
Remark 7.4.8 The existence of the character x is not known in general, but there are standard conjectures in Iwasawa theory which would guarantee that such char- acters exist Some results in this direction (although not applicable to the case considered here) have been obtained by Vatsal in [Vat03].
Hecke algebras and deformation rings 000008 123
Congruences and weak congruences 1 0 LH HQ HQ ee eee 123
Let £ be an odd prime Recall that we have fixed embeddings Q => Q, C and let
E denote a finite extension of Q, containing all Hecke eigenvalues of all the eigenforms in Sp_1 (4, (=4)) Let O be the ring of integers of E with uniformizer À, and put
F =Ó/^ Let N be the basis of 5y_¡ (4, (S)) consisting of normalized eigenforms. The elements of NV can be written as fi, fƒ, ; fr: fP, fz+i, fs, where fy # f? for 12 ¡ a(n)q", g = Oe, b(n)q” EN, and f =„ g, then either ƒ = g orf = g0.
Proof Assume ƒ =„ g Then we have À; = Xp For p # 2, £, we have tr p(Frob,) = a(p) and tr p,(Frob,) = b(p) Since f =, g, this means that tr p;(Frob,) = tr p,(Frob,) for p split, and tr p;(Frob,) = tr ỉ;(robi) = tr7,(Erob2) = tr p,(Frob,) for p inert, where p = pOx Thus the traces of p, and p, agree on the set of Frobenius automorphisms at all the primes of K except possibly at the primes over 2 and £. Also as det p(Frob,) = (=) p’? = det pg(Frob,), we have (by the Tchebotarev Density Theorem and the Brauer-Nesbitt Theorem) Pilcx = Plex By possibly changing a basis of, say, ỉ„, we may assume that ỉ†\œ„ = Py lex We will now show that this implies that ỉ‡? and 7ỉ2° differ at most by a twist by x, where x is the Galois character associated with the Dirichlet character (=*) Indeed, let o € Gg and let c € Gag denote the complex conjugation. Then ứ?(ứ) = p;°(c) ifo € Ge Assume o  Gx Then there exists a unique ơ € Gr such that ơ = co’ We have
7ÿ (co!) = By (e)P† (ứ') = BF (e)py'(e)""ụ'(e)pÿ'(ứ') = BF (CIPS (C)"75(0).
De(c)Py (c) Pe (o) ifo Gx
As c is of order 2, p4°(c)ps°(c)"' = 1 or —1 In the first case, Z‡# = 5°, and in the latter case 2# = yp Hence a(p) = (=4)' b(p) for some i and all p # 2,ý Now, if £ is split in K, then a(@) = b(0) If on the other hand ý is inert, then we must have a(@)? = b(ỉ? Assume that @ is inert If a(@) = 0, then (2)? = 0, hence b() = 0, ie., ƒ and g are congruent at ý If ứ(9) # 0, then we must also have uy *% 1 b(Ê) # 0, hence both ƒ and g are /-ordinary In such case ỉ/|p, = f with ty unramifed and ?(FTob¿) is the unit root af of X?~a(£)X + (=) #-2 (ch Theorem d3
10.1.2) Analogous statements hold for p, Now, since p; = 6, ® x’, we must have u?(ERrob¿) = (=4)' 12 (Frob,) (mod \), hence ay = (=!) ay As ay is the unique unit root of the polynomial X? — a(6)X + (+4) é*-?, we must have a(¢) = ay, and similarly b(£) = ag, hence we conclude that a(0) = (#) b(£) Thus we have ƒ =,2} g or ƒ =f} g? Then by Lemma 8.1.9 f = g or f = g? O
Corollary 8.1.11 [f f,gEN,gI?, and ƒ =y g, then f = g.
Proof Since f =ằ g, we must have f = g or f = g? by Proposition 8.1.10, but gI° Oo
Let f,g € ẤM By Proposition 8.1.10, if g =, f, then g = ƒ or g = f? Hence if f = f?, we have my = my = mye If on the other hand ƒ # f?, then mạ = my or mạ = myo This proves the following corollary.
Corollary 8.1.12 Let m; = ker dy € MaxSpecT’ for feN If f = ƒ°, then
M(m;) = {my} If f # ƒ°, then M(m;) = {my,myo} Hence, if f = f’, we have an injection Tw, > Tm,, while if f # ƒ°, we have Thy, = Tay X Tyo
Let g = 77, b(n)q” € N with g = g? and g =y f = ~~, a(n)q" Then by Proposition 8.1.10, g = ƒ or g = f? However, as g = g? we must have b(p) b(p) = —b(p), i.e., b(p) = 0 for all inert primes p Hence f = g would imply that a(p) = 0 for all inert p As a(p) = —a(p) for such p, we must have a(p) = 0, i.e., a(p) = b(p) = 0 Moreover f =, g implies a(p) = b(p) for all p split in K, and since we have a(p) = a(p) for such p, we get a(p) = b(p) for all p # 2, i.e., g =tay f?, hence by Lemma 8.1.9, g = f°, which implies f = f° Thus if ƒ # f°, there are no forms g € N, 9 = 9? which are weakly congruent to either ƒ or f?.
Proposition 8.1.13 7ƒ ƒ € N, then the canonical O-algebra map óo : Ty, — Tn, 1S injective.
Proof lf f = f?, then Ty, injects into Tạ, by Corollary 8.1.12 Assume that f # f? Then, as remarked before, g =, f implies g # ứ° By Proposition 8.1.10, g9 = f or g = ƒ° Without loss of generality we may assume that g = f As g° =w g, we have g? =, f, hence again by Proposition 8.1.10, g° = f or g? = f?.
However, ứ” = f would imply that g? = g and we saw that this is impossible, so g? = ƒ? Thus if we consider Ty, as a subalgebra of Len o=f O, where the embedding Tn, => [eager O is given by T +> (À;(T));, we can also consider
Tự, => [gen gay O where T > (Age(T))g By Corollary 8.1.12 we have Tw => Tiny X Tiny: so we just need to prove that the composite Tw, — Tn, X Lingo — Tn, is injective, where the last arrow is projection Identifying Tm, with a subalgebra of [gen gap O and Tm, x Typ with a subalgebra of [[jey gap? X gen gaz OP by the embeddings specified above, we see that T € Tw, maps to an element of Hoe ỉ x Ten gar â, whose g-entry in the first product is the same as the corresponding g-entry in the second product for every g € NV, g = ƒ, since Tg = ag implies Tg? = ag® for T € Ty, Hence if J maps to zero under the composite
Tw, — Tiny X T¿¿ 7 Tm, it must be zero in Tạ, X Tm,,, which means T’ = 0 in
Ty Hence the map Tw, — Tm, is injective Oo
Deformations of Galois representations 2 0 ee 130
The goal of this section is to prove surjectivity of do : Tw, — Tm, We will use the theory of deformations of Galois representations For an introduction to the subject see e.g [Maz97].
Let C denote the category of local, complete O-algebras with residue field F and let G be a profinite group A morphism between two objects in C is a continuous O- algebra homomorphism which induces the identity on the residue fields For an object
R of C we denote by mg its maximal ideal Consider a continuous representation p:G— GL2(F) If R is an object of C, a continuous representation ứ : G — GLa(R) is called a deformation of p if p = p modmr Two representations ứ and p’ are called strictly equivalent if p(g) = xp'(g)x~ for every g € G with z € 1+ Mo(mp)
131 independent of g We will write ứ = p’ if they are strictly equivalent A pair (RTMY, op) consisting of an object R°” of C and a deformation ứ*2Y* : G — GLa(R) is called a universal couple if for every deformation p : G — GLa(), where F is an object in C, there exists a unique O-algebra homomorphism ở : RY” —, R such that óo pTM’ = pin GLạ(R) The ring R°TM” is called the universal deformation ring of p By the universal property stated above, it is unique if it exists Note that any O- algebra homomorphism between objects in C is automatically local, since all objects of C have the same residue fields [Indeed, if S and T are objects inC, and ý: S — 7 is an O-algebra homomorphism, then we have F = $/mg ô S/W7 (mrp) => T/mp =
F Hence S/#7!(mr) = F and wy} (mr) = ms.]
Theorem 8.2.1 (Mazur) Suppose that p: G — GL, (F) is absolutely irreducible. Then there exists a universal deformation ring RY” inC and a universal deformation un\v : G — GL,(RTM).
Proof [Hid00], Theorem 2.26 with a remark that for a Noetherian ring the notions of pro-artinian and complete coincide Oo 8.2.2 Hecke algebras as quotients of deformation rings
Consider ƒ € N and let pr : Ga — GLa(Ó) be the associated Galois representa- tion (after fixing a lattice in #”) Let ỉứ;: Gq —> GLa(F) be its reduction modulo
Xr Since pz is unramified away from S = {2, ¢}, it factors through Gas, the Galois group of the maximal Galois extension of Q unramified away from S Let Gx,s be the image of Gx under the map Gx ~ Gq — Gag Then we have a commutative diagram
0——>Œy„g——>Gq,g—— Gal(K/Q)——>0 in which the rows are exact We will be considering deformations of the represenation
Pr : Gas — Gho(F) and py x := Prleg.s- From now on we assume that both 9; and ỉ; „ are absolutely irreducible Let (Rg, pg) and (Rx, px) denote the universal couples of ứ; and py x respectively, which exist by Theorem 8.2.1 We will denote
Mr and mr, by mg and mx respectively Let A be a density zero set of primes of Q and g € N, g =a f Then py : Gas —> GLe(Q) is a deformation of ỉ;, and pglc, is a deformation of Ps x (this follows from Lemma 8.1.7) As in the proof of Proposition 8.1.13 we identify Tm, and Tw, with appropriate subalgebras of ]] ge, gap O and of Hye N g=uf Oằ Tespectively Let T denote the O-subalgebra of T generated by the operators T,, for p # 2, @ and let T’ denote the O-subalgebra of T’ generated by the operators 7; (p split p # £) and by 772 (p inert, p 4 £) We put My := Tn my and tị := T n my Let Ly denote the subset of NV consisting of those eigenforms which are congruent to ƒ except possibly at 2 or ý Similarly let 2; be the subset of N consisting of those eigenforms which are weakly congruent to f except possibly at 2 or ý We have 3; C & Note that we can identify T,, (resp.
Tạ, ) with a subalgebra of ll;s>, Ó (resp Ilex O) via the map T; +> (b,(p)) gen;
(resp T, (0;(p));ex,), where g = yr bg(n)g” Consider the representations
0:= ll;s>, Pg: Gas > GL (Tycx, 0), and ỉ := ứ|e„,s Choose bases for each p, so that 6, = py for all g, g’ € Uy, and so that p,(c) = [! _¡] for all g € >;, where c is the complex conjugation.
Lemma 8.2.2 The image of the representation p is contained in GL2(Ta,).
We claim that ứ'(Œz,s) is contained in the image of GLạ(T, ) inside GLa(Ta,) raf my
To prove it, let @ denote the map Ty — Ta ; induced by TT In fact, if we
133 identify Ta, with a subalgebra of [] O and Tạ, with a subalgebra of [] 962, O, g€5/ then  is just the restriction to Tạ, of the natural projection Toes, O -ằ In O.
It is easy to see that o(Liy ) is local It is clearly complete and Noetherian, and its residue field is F Hence o(Tyy) is an object of C Consider pf’ : Ớxys — GLằ (Tye, 0), ð(ỉ) = (04(2) ges, è.e., we have gop! = ỉ' For 7 € Gx,9 we denote by [7] the conjugacy class of 7 in Gx.g Note that Œx,s is topologically generated by the set |J;csseo„,pazzsltFrobpl Moreover, if p is inert, then Frobp = Frobs, where p = pOx, while if p = pp is split, then the choice of the prime $8 of the ring of integers of the maximal Galois extension of Q unramified away from Š which lies over p, determines whether Frob, = Frob, or Frob, = Frobg In any case, switching from one choice to the other results in conjugating the fixed decomposition group of the prime ?# by an element of Go s, which doesn’t alter the trace of p’. Hence for a split p, we have trp’(Frob,) = trp’(Frobp) = trp’(Frob,) Note that for p split trứ(Frob,) = Tp € Tạ, and for p inert trứ(Erob2) = (trứ'Œtrob,g))? + det p/(Frob,) = T? — p*-? € Tạ, Thus trỉ(Gx,s) C Tạ, which implies that trứ(Gx,s) = tr (doZ(Gr,s)) C ú(T, ) By a theorem of Mazur ([Maz97], Corollary
6, page 256)) this implies that (after possibly changing the basis of p’), we have ỉứ{GŒx,s) C GLa(2(T,)) Then by Lemma 4.1 in [dSL97], ứ is a deformation of py and ỉ : Ges — GLo(o(Tiy, )) is a deformation of ỉ;„ Hence there are unique O-algebra homomorphisms ¢g : Rg — Tạ, and éx : ly — d(T), such that
69° pg © p, and ox o px & p’ In fact as pale, is a deformation of p; x, there is a unique O-algebra homomorphism Ú : Rx —› Rg, such that Wo pr % pale,- Hence we get the following diagram
(8.3) Re—— Ro o(Tyy) —— Tạ, where ¿ denotes the embedding o(Tyy )C Tạ As the composite
10 p': Gx — GLo(o(Ty)) + GLo(Ta,) m is a deformation of ỉ; x, there is a unique O-algebra homomorphism a: Ry —> Tạ, " such that œòy © Lop’ Since dx opK & p' we get LodgopK & Lop’, hence ,oĨy =a by uniqueness of a On the other hand as stated in the paragraph before diagram (8.3), Wo px & pale,x, thus đo 9 9 pK óo 9 pala, Since dg 9 pg & p, we have
$0 9 Polex â Pla, = +40 ứ Hence Âg° Wo px 140’, which implies as before that óo 9Ú =a So, uoóy = óoo0, and thus the above diagram commutes Moreover as tr p(Frob,) = Tp for p # 2, and óo(tr pg(Frob,)) = tr (¢gopq(Frob,)) = tr p(Frob,), we see that all T, for p # 2, £ are in the image of Rg So, óo is surjective Similarly óx is surjective Our goal now is to prove surjectivity of ý which will imply surjectivity of ¿.
The map ý : Rx — Rg is local (by the virtue of residue fields of Rg and Rx being the same), hence induces an O/A-linear homomorphism on the cotangent spaces mx /(mz,ARK) > mọ/(mỗ, Rg), which we will call woe (Note that as O > Rx is local, the image of À is contained in mx; however, the ideal ARx Z mx Similarly for Rg.) Let C denote the cokernel of /„ : ma /(mz,ARK) — mọ/(m,ÀRq) As the above map is just a map of F-vector spaces, the sequence of dual maps is exact:
0 — Homp(C, F) + Homp(mg/(m3,, ARg), F) “+ Homp(mx/(m%, ARK), F).
135 where 5: œ >œo„; We will prove that Wi, is injective, which will imply that C=0.
8.2.3 Zariski tangent space of the deformation functor
For a profinite group G and the universal couple (R¥’, op’) of a representation ứ:G — GL;(F), there is an action of G on the space of 2x2 matrices with entries in
F given by g-M = p(g)Mp(g)7! This G-module will be denoted ad(5) We will write
R instead of RY to shorten notation The F-vector space Homg(mz/(mậ, AR), F) is called the Zariski tangent space of the deformation functor representaed by R.
Lemma 8.2.3 If 7 is absolutely irreducible, then
Homg(ma/(m, \R), F) % H*(G,ad(p)), where the cohomology group stands for continuous group cohomology and ad(p) is a discrete G-module.
When G = Gq (or G = Gx) and R = Rg (or R = Rx), we will denote the isomorphism from Lemma 8.2.3 by tg (or tx, respectively).
Proposition 8.2.4 The following diagram is commutative:
(8.4) Homr(mo/(mộ, \Rg), F) —2—> Homp(mx/(m3,, ARx), F)
Proof We first unravel the isomorphism tr from Lemma 8.2.3 There is a natural sequence of bijections:
Homp(mp/(m, AR), F) % Homp_aig(B/(m2, AR), Fle]) Homo_az(R, Fle]), where Fle} denotes the dual numbers To explain this we first note that the exact sequence of F-vector spaces
0 — mg/(m2 + AR) > R/(m, +R) > F- 0 has a natural splitting sp : F #/(mệ + ÀR) coming from the structure morphism
O — R This provides us with an isomorphism of F-vector spaces
Let pr denote the natural epimophism R/(m? + AR) -ằ R/mpg = F Then the map (8.6) is given by (pr,idp — so pr) The inverse to the map (8.6) is given by sz +i, where i is the embedding of mg/(m2 + AR) into R/(mộ +ÀR) To see that this map is the inverse of (8.6), one only needs to show that the composite ể@nosg: F —ơ R/(m2,+AR) — F is the identity This follows from the commutativity of the following diagram © ———*> R/(m2, + AR)
The upper triangle is clearly commutative, and so is the square since all maps in C induce identity on the residue fields This implies that the lower triangle commutes as well, which is the asserted claim.
The first bijection in the sequence (8.5) comes from the F-vector space decom- position (8.6) Indeed, any O-algebra map from R to Fle] must factor through R/(m2,ÀAR) Also, any such map being an O-algebra map must send F in the de- composition R/(m?,, \R) = F@ma/(mậ, AR) to F in the decomposition Fe] = F@Fe via the identity map, and being local it must send the maximal ideal part into the
Hida’s congruence modules 2 ee 144 9 Automorphic representations 2 0 ee 149
We will now introduce (following Hida [Hid87]) the notion of a congruence module and a Hida invariant Let A be a domain of characteristic zero with fraction field L. Let R be a finite flat A-algebra and let y: R — L be an A-algebra homomorphism.
We assume the following splitting condition: @ induces an algebra decomposition R®,s,L = L®B, with a complementary direct summand Bz, i.e., y is the projection to the first factor Let B denote the image of R under the composite R~ J.@ Ù —
Br, where the second map is projection Call 6 the diagonal embedding of R into A@®B Then the congruence module of y is defined to be the cokernel of 6.
Lemma 8.3.1 Let L be a number field and L'/L a finite extension Let A and A’
145 denote the rings of integers of L and LrespectiuelU If @ : R — A is an A-algebra homomorphism satisfying the splitting condition, then p@id: R@,4A' — A’ satisfies the splitting condition and R®, A’ is A’-flat Moreover the congruence module of œ@id is isomorphic to the congruence module of y tensored over A with A’, and if L’ is sufficiently large, there exists n € A’, n #0, such that the congruence module of ~ ® id is isomorphic to A'/nA’.
Definition 8.3.2 The element 7 in Lemma 8.3.1 will be called the Hida invariant of R.
Fix f € Set Np := {ge |m¿ = m/}, A= Ó, R= Ty, and L = E Since Sha (4, (=)) has a basis VV of newforms, Tụ, C Ì sex; Ó is a finite flat O-algebra. The map []jey, Ag! Tm, > [jen O induces an E-algebra isomorphism Tụ, @ E Tye Ns E It follows that A, satisfies the splitting condition, the complementary direct summand being lv E Let B denote the image of T,,, under the composite Tn, => Tn, @ È — Leach E, and 6 the embedding of Tạ, into O@®B Then enlarging # if necessary, we obtain (by Lemma 8.3.1) the isomorphism coker 6 = O/nO for some 7 € O This 7 is by definition the Hida invariant of Tm,.
We fix it in what follows Set V7 := {g € N | my = m4}.
Proposition 8.3.3 Assume f € N is ordinary at £ and the associated Galois repre- sentation ps is such that Pr|œ„ is absolutely irreducible Then there exists T € Tw, such that Tf = nf, Tf? = nƒ° and Tg = 0 for all g € N; \ {f, f?}.
Proof First note that Ty, can be identified with the image of T” inside Ende (Sx_1, 5),where Sy—1,¢ C Sp—1 (4, (=4)) is the subspace spanned by NV} By Corollary 8.2.9,the natural O-algebra map Tw, — Tm, is an isomorphism So, it is enough to find
T € Tm, such that Tf = nf and Tg = 0 for every g € Ny \ {f} (Note that by
Proposition 8.2.10, f? Â Ny.) We have a short exact sequence 0 — Tm, >ỉ@B—ơ
O/nO — 0 Consider an element (7,0) € O@ B This element is mapped to zero under O @ B — O/nO, hence is in the image of Tụ, ~ O @ B Let T be the preimage of (7,0) under this injection This T has the desired property L]
Our goal now is to relate 7 to (ƒ, ƒ) and to the “integral” periods OF, QF (cf. [Vat99]), which we define below.
Let R be a ring and N a positive integer Define L,_1(R) to be the R-module of homogeneous polynomials in 2 variables (X and Y, say) of degree k — 3 The semi-group GL2(Q)t MN M2(Z) acts on Ly_1(R) via
We let #2(T:(N), Lp_-i(R)) denote the parabolic cohomology with coefficients in Ly-i(R) (For definition and basic properties of parabolic cohomology, see e.g. [Hid93], Appendix.) It is a Hecke module (with the action defined by (8.12)) For ƒ€ 6® :(T¡(N)) put wy = f(z)(2X+Y)*oldz Let zo € H, where H is the complex upper half-plane The map ô(ƒ) : y > f[/ wy defines a cocycle whose cohomol- ogy class inside H2(T¡(N), Ly-1(C)) is independent of zo In fact 6(f) defines an isomorphism
6(f) : Se-i(Ti(), C) @ Sp-i(Ti(N), C) > Hp(Ti(N), Le-1(C)) called the Eichler-Shimura isomorphism.
Let HB(T1(N), Le-1(C))* (resp Hp(Ti(N), Lp-1(C))~) denote the subspace ofHẸ(T1(N), Ly-i(C)) on which the complex conjugation acts by +1 (resp by -1) and define %_¡(T¡(V),C)T and S,_1([1(N),C)~ similarly Our aim is to investigate
147 possible “integral” structures on the space of modular forms There are two obvious choices: one is the O-submodule of S„_+(T1(N), C) consisting of forms whose Fourier coefficients at co he in Ó Another one is the O-submodule
The periods OF and 9; provide a connection between these two structures.
Assume f is a newform Consider the O-algebra map T %1, O, where À;(7) is the eigenvalue of the operator T on the eigenform ƒ Let py := ker À; and my be the kernel of the map T 4,040 /A, where the second arrow is the reduction mod À.
Assume now that the Galois representation 7ỉ; is irreducible This ensures that the induced map
Hp(Ti(N), Le-1(O))m, — HE(TI(M), Le-1(EZ)) mj is injective Let Hp(T1(N), Le-1(O))in, [Py] be the O-submodule of
Hp(Ti(N), Le-1(O))in, annihilated by py We have Hp((1(N), Le-1(R))n, [ps] % R for ? = +,—, where R= 0O,E,C since ƒ is a newform and # by assumption contains all the Hecke eigenvalues of f Let 5; be an O-generator of Hp(Ti(N), Le-1() ing [py] There exists a unique element 6(f)’ € Hp(Ti(N), Le-i(C))in, [Pl for ? = +,— such that
The quantities oO are called the integral periods of f and the freedom of choice of the generators 6; is reflected in the fact that ; are defined only up to an O-unit.Proposition 8.3.4 ([Hid87], Theorem 2.5) If ƒ is a newform ordinary at £, then where (%*) is a A-adic unit if Ê A 2.
Satake parameters
Satake parameters are a way to parametrize unramified representations of p-adic algebraic groups Placed in a more general context they can be regarded as a special kind of Langlands parameters, which we define below As automorphic representa- tions are unramified at almost all places, Satake parameters are a useful tool in the study of automorphic forms.
In this section G’ will denote a quasi-split reductive algebraic group defined over a local or a global field of characteristic zero In what follows E will denote a local field and L a number field For a place p of Ù we denote by L, the completion of L at p, and for every p we fix an embedding Ù — Ly.
Definition 9.1.1 Suppose G’ is defined over E Let V be a C-vector space A continuous representation 7 : G’(E) — GLc(W) is called admissible if for every open compact subgroup K’ of G’(E) the space V*" of K’-fixed vectors is finite dimensional.
We will be interested in parametrizing the set of isomorphism classes of irre- ducible admissible representations of G’(E) We begin by defining the (in many cases only conjectural) Langlands parameters and then formulate the theory of Sa- take parametrization as a special (known) case of the Local Langlands Conjecture.
There is a short exact sequence
1——+ I —> Gal(E/E) > Gal(R/®) — 1, where J denotes the inertia subgroup of Gal(£/E) and x the residue field of E. The group Gal(K/ô) has a canonical generator, the Frobenius element Frob which furnishes a canonical isomorphism Gal(E/g) > Z given by Frob + 1 Here 2 :=
IS Z, We define the Weil group We of E to be y71(Z).
Definition 9.1.2 The L-group of G’ is the semi-direct product G’ x Gal(E/E), where G’ is the complex dual of G’ (i.e., a complex Lie group whose root datum is dual to that of G’) For the definition of the action of Gal(E/E) on G’ see for example [BCdS703], page 253.
Conjecture 9.1.3 (Local Langlands Conjecture) There exists a surjective map with finite fibers from the set of isomorphism classes of (non-special) irreducible admissible representations of G'(E) into the set of semi-simple G'-conjugacy classes of homomorphisms Wz —› G' x Gal(E/E), where we impose a compatibility condition and demand that the projection of the image of Wz onto Gal(E/E) be the natural embedding Wg = Gal(E/E).
Definition 9.1.4 Let G’ be such that the Local Langlands Conjecture holds for it The semi-simple G’-conjugacy class of homomorphisms Wg — G’ x Gal(E/E) corresponding to an irreducible admissible representation 7 of G’(E) is called the Langlands parameter of 7.
Remark 9.1.5 The full version of the Local Langlands Conjecture includes all irreducible admissible representations of G’(F), but to capture the “special” ones Wz needs to be replaced with a variant called the Weil-Deligne group For a definition of
151 the Weil-Deligne group see [Tat79] and for a description of special representations in case G’ = GL, see e.g [Gel97] We will not need precise definitions of these notions here, but only note that all representations we will consider will be “non-special” in that sense.
Remark 9.1.6 The Local Langlands Conjecture is usually stated in a much more refined form, describing for example which Langlands parameters correspond to tem- pered, square-integrable, etc representations See, e.g [BCdS*03], section 11).
Remark 9.1.7 The Galois group Gal(#/E) in Definition 9.1.4 can be replaced with Gal(E’/E), where # is any finite unramified extension of E over which G’ splits.
We fix a certain open compact subgroup Ay, C Œ(#) called the Iwahori subgroup.
We will not give its definition here (see, e.g., [Bum97]) and only note that in the cases we will need it, i.e., when E = Q, and G’ = GL, or G’ = U(2,2), we have
Definition 9.1.8 Let 7 be an irreducible admissible representation of G’(E) and V the space of 7 The representation 7 is called unramified, if there exists v € V fixed by Kw.
Unramified representations (important in their own right) are extremely useful in the theory of automorphic forms In fact, let 7 = &; Tp) be an automorphic representation of G'(A;) The factors 7; are irreducible admissible representations of G'(L,) Moreover, 7p is unramified for all but finitely many p.
Example 9.1.9 Let ƒ € Sn(N,w) be a newform, and let z = @,7/; be the corresponding automorphic representation of GL2(A) The representation 7,ằ is unramifed if and only if p f Noo
Example 9.1.10 An analogue of Example 9.1.9 is true for hermitian modular forms.
In particular, if F’ is a hermitian modular form of full level, then the corresponding automorphic representation II] = 6; lÍz:; is unramified at every finite p (i.e., Ip, is unramified for every finite p).
In the case of unramified representations the Local Langlands Conjecture is a theorem due to Satake et al.
Theorem 9.1.11 (Satake) There exists a one-to-one correspondence between the set of isomorphism classes of unramified representations of G’(E) and the set of semisimple G'-conjugacy classes of homomorphisms Wg — G’ x Gal(E/E) trivial on ẽ and satisfiying the same compatibility condition as in Conjecture 9.1.3.
In fact in this case the correspondence is explicit, as we explain below First note that since the Langlands parameters for unramified representations are trivial on J, they are completely determined by their value on the Frobenius element Moreover, a semi-simple conjugacy class of homomorphisms has a representative whose image is contained in T’ x Gal(E/E), where T’ denotes the diagonal torus of G’ For an unramified representation 7 we denote a (non-unique) choice of this representative by 7.
Let B’ = T’'U’ be a Borel subgroup of G’ with 7” a maximal torus and U’ the unipotent radical It is a standard fact that every unramified representation of G’(E) is obtained from a continuous character x : T(E) — C* by first extending it trivially to U'(E) and then inducing it to G’(£) If 7 is an unramified representation induced from a character y : T(E) — C* then 7(Frob) = diag(fi,tfa, , f„) x Frob.
Definition 9.1.12 The complex numbers {f,fa, ,f„} are called the Satake pa- rameters of 7 We point out that the set {t1,t2, ,t,} itself is not unique, but the
153 Œ-conjugacy class of the element 7#(Frob) = diag(fi,fa, ,É„) x Frob is.
Maass lifts revisited 2 ee 154
In section 4.1 we defined the Maass lift
Sk-1 (ô (=)) — &(fz) fro Fy,—4 and in section 6.2 (Proposition 6.2.3) we showed that
(9.2) Ly (s, F,x) = L(BC(ƒ), s —2+k/2, wx) L(BC(ƒ), s —3+k/2, wy).
We now rewrite (9.2) using Dg(s, 7, x) Let f= an)” € Sis (4 (=)) n=l
155 be a normalized eigenform with f? # ƒ and let ; = ®, Typ be the corresponding automorphic representation of GLạ(A) As usually, we fix an embedding Q => Q, for every p As noted in Example 9.1.9, 77, is unramified for p # 2,00 Moreover, the representation Bz/q(Z;) = &, BCx,/a,(7F,) of (Resx/q GLs)(Ä) is unramified at all finite places p of Q The L-group of Resx/q GL¿ is
(GL2(C) x GL,(C)) x Gal(Q/Q), and the action of Gal(Q/Q) on GL2(C) x GL2(C) factors through Gal(K/Q), where the action of the non-trivial element o € Gal(/Q) on GLa(C) x GL2(C) is given by a, ỉ8) > (B,a) For p Ơ 2, let a, af? denote the p-Satake parameters of 7ằ Wep,1! “"p,2 f use the superscript ‘rep’ to emphasize that these are the ‘representation-theoretic’ Satake parameters as opposed to the classicial Satake parameters Q)1, Qp.o of ƒ rep rep defined in section 4.2 In fact, Œ„}, Œp3 are related to @p, and dp For p # 2 one has
Then the GL2(C) x GL2(C)-conjugacy class corresponding to BCx,/@, (mfp) is given by rep rep
As BCz/q(Z;) is also unramified at the prime 2, it makes sense to talk about its Satake parameters there In fact for p = 2, the conjugacy class corresponding to BCx,/q,(Tf,p) also has the form rep rep pl pl: x Erob›, rep p,2 œ rep p,2 where ai =2! #/2a(2), (9.4) ane !ˆ°/a().
Note here that for p # 2, one has re —4 rep\—1 —4 “rep ont = (2) = (5) a
This follows from the fact that ap1+ Qp2 = a(p) and ap1dp.2 = (=*) p*-? together with the formula a(p) = (=) a(p) (Fact 2.3.1) [Indeed, ap; + Qp2 € R if p Pp splits and Qj); + Qp2 € iR if p is inert So, Ima,; = —Imap, if p splits and Redpx = —Readp if p is inert Moreover, as œpid,› € R for all p z# 2, we get
Re ap Ím a@p2 = —Re œp2ẽm api, hence for p split we get Reap = Reap.ằ and for p inert, Imay1 = ma So, œp¡ = Apa if p splits and ap = —Qp2 if p is inert.] Let Ir, = @, Ilr, be the automorphic representation corresponding to the eigenform Fy As pointed out in Example 9.1.15, the representations Ip,ằ are unramified for every finite p Let p be any finite rational prime and denote by Àj1› -› Apa the p-Satake parameters of IIp, The relations (6.6) and (6.9) translate into
(9.5) hea = Pita" (ase for 7 = 1,2, where p is the prime of K over p determined by the choice of embedding
So we obtain where L(s, BCz/g(T/),Ú) := L(s,BCx/g(x/) @ Ú,2) with + : GL¿(C) — GLa(C)
(A91 and ~ any Hecke character of Aj, regarded as a character of (Resx/q GLi)(A) Note that while the set {A„T, , À;a} depends on the choice of the embeddings Q, c Q,for p split, the L-function L.(s, lĨr,, x) is independent of that choice.
CAP representations for general imaginary quadratic fields
We now define CAP representations in general Let G’ be a quasi-split reductive algebraic group defined over a number field L Let P’ = M’U’ be a maximal parabolic subgroup with Levi M’ and unipotent radical U’.
Definition 9.3.1 Let I = @, I, be a cuspidal automorphic representation of G{(Ar) We say that II is CAP (cuspidal but associated to a parabolic subgroup) associated to P’ (or simply CAP if P’ is clear from the context) if there exists a cuspidal automorphic representation 7 = ®), 7p of M’(Az) such that I, = 3y for almost every prime p of L Here © = É, Uy is the Langlands quotient of Ind p42} T, where we again extend 7 trivially to U'(Az) We also sometimes say that IT is CAP associated with 3 at p if II, = Xp.
From now on let G’ denote the group U(2, 2) associated with an imaginary quadratic extension L/Q of discriminant D; Let P’ be the Siegel parabolic (defined in the same way as P for the case L = K, cf section 2.2) and
(4) its Levi subgroup Let k be a positive integer with #ÓƑ | k Let ƒ be a newform of weight k — 1, level 2y and character being the quadratic character associated with the extension L/Q (cf [Kri91]) Let z; = Ó9, 7 be the corresponding automorphic representation of GL2(A) We fix an isomorphism
Then we can regard BC;/q(Z¿) as an automorphic representation of M‘(A) Extend BC;/q(7;) trivially to the unipotent radical of P’(A) We will denote this extension by BCr/q(Z/) as well Set
Let Ly denote the Langlands quotient of > ;- The representation Uy is automorphic and occurs inside the residual spectrum of G’(A) (cf [KN98], Theorem 1.1) Note that det G(A) = Aj, := {am € AE | a € Af} Here (-) denotes the action of
Let w: Ary — C* be a unitary character unramified at all finite places Set
The representation Ly, is an automorphic representation of G’(A) unramified at all finite places.
Conjecture 9.3.2 There exists a CAP representation Uy = &, pw,ằ of GA) associated with Uy, at all finite places.
Note that such a representation II;,, is automatically unramified at all finite places and its Satake parameters are equal to those of Up
Example 9.3.3 If = K, f € Sp_1 (4, (=*)) and w is the unique Hecke character of K unramified at all finite places, whose oo-component equals (z/Z)~*/? then II fw
159 exists and equals the automorphic representation corresponding to F;, the Maass lift of f defined in section 4.1.
Assuming Conjecture 9.3.2 it is not difficult to relate the Satake parameters of
II and the Satake parameters of BC;/o(7Z/) In fact, for any Hecke character x: AZ — C* one has ủ#(s, Hy, x) = L(s — 1/2, BCxa(ts), wx) L(s + 1/2, BCkq(®/), wx).
Galois representations and Selmer groups 0+.0004 160
Galois representations 2 1 Hà gà va na 160
In this section we gather some facts about Galois representations attached to elliptic and hermitian modular forms.
We begin with some generalities Let L be a number field and set Œr := Gal(L/L). Let £ be a rational prime and # a finite extension of Q¿ with ring of integers O and a uniformizer A Set F := O/A Let V be a finite dimensional E-vector space on which Gy; acts continuously This action defines a continuous homomorphism
0: Gr —> GLg(V), which we will call an (€-adic) Galois representation As Gy is compact, there exists a Œr-stable O-lattice T C V, so p gives rise to a representation pr : Gz — GLo(T). Continuity of p implies that p(g) maps AT into AT for every g € G, so pr descends to a homomorphism ỉy : Œy — GLp(T/A), which in the literature is sometimes refered to as a residual Galois representation We will call ứy the (mod A) Galois representation associated with T or (more abusively) the (mod @) Galois representa- tion associated with T For g € GL(V), gTg~ is another G;-stable O-lattice and the mod À representation Dyr,-1 : Gp — GLe(gTg~'/X) may not be isomorhic to fr.
However, pr and ỉ„r;-: have isomorphic semi-simplifications pr = Py7,-1, hence we
161 can drop the subscript indicating the choice of the lattice and refer to the (mod À) semi-simplification of p as a well-defined object p* (defined up to an isomorphism).
It follows that if Bp is irreducible (resp reducible) for some T, then it is so for all
T Hence for an irreducible p; it makes sense to talk about p without refering to a concrete lattice.
Theorem 10.1.1 (Brauer-Nessbitt) Let G be a group, R a field and V a finite dimensional R-vector space Two representations pi, 0s : Gr — GLr(V) have iso- morphic semi-simplifications if and only if the characteristic polynomials of p\(g) and pa(g) coincide for every g EG.
For every prime p of L we fix an embedding L — Ly This singles out a unique prime p of L over p Denote by D, C G, the decomposition group of ÿ and by
I, C Dy, the inertia group The embedding L Ly allows us to identify D, with Gal(L,/L,) Let k, denote the residue field of L, There is a canonical isomorphism Gal(K,/ky) = D,/I, and hence a short exact sequence
Let ƒ = }}¡ a(n)q" € Sm(N,) be a newform Let Ly be a finite extension of
Q containing the Hecke eigenvalues of ƒ.
Theorem 10.1.2 (Deligne, et al., [Hid00], Theorem 3.26) Let E be a finite extension of Qe containing L; There exists an irreducible 2-dimensional Galois representation pp : GŒq —> GLe(V) unramified away from primes dividing N£ and such that
1 If p { N2, then the characteristic polynomial of ps(Frob,) is given by X? — a(p)X + p(p)pTM?.
2 Ifm>2, £4 N, and f is ordinary at É, ¡.e., a(@) ts an Ê-adic unit, then
Li where dự, ps are characters with ty unramified and such that u‡(Erob,) 1s the unique unit root of X2 — a(ÐX + (0£ ®—!, In particular one has
1 where € denotes the €-adic cyclotomic character.
3 Let p{ £ be a rational prime number Let p° be the largest power of p dividing
N, and p° be the largest power of p dividing cond If e = e > 0, then ự
1 where wy’ is the Galois character associated to w by class field theory.
We now consider the case of hermitian modular forms Let ~ be a Hecke character of K, and F € S,,(N,w) a hermitian modular eigenform By Theorem 5.2.14 there exists a finite extension Lr of Q containing the Hecke eigenvalues of F’.
Theorem 10.1.3 There exists a finite extension Er of Q¿ containing Lr and a 4- dimensional semisimple Galois representation pr: Gx + GLeg,(V) unramified away from the primes of K dividing 2N£ and such that
(i) For any prime p of K such that p { 2N@, the set of eigenvalues of pr(Frob,) coincides with the set of the Galois-Satake parameters of F at p (cf Definition5.9.9):
(i) If £4 N, and p is a place of K over £, the representation pr|p, 1s crystalline (cf section 10.2).
(iii) If€ > m, &{ N, and p is a place of K over £, the representation pr|p, is short. (For a definition of short refer to [DF G04], section 1.1.2.)
Remark 10.1.4 We know of no reference in the existing literature for the proof of this theorem, although it is widely regarded as a known result For some discussion regarding Galois representations attached to hermitian modular forms, see [BR94].
We assume Theorem 10.1.3 in what follows.
Let k be a positive integer divisible by 4, and ƒ € S,_ (4, (=*)) a normalized eigenform Assume that the mod £ Galois representation fy attached to f is such that ỉ;|ứ„ is irreducible (and hence ứ; is well-defined up to an isomorphism) Then f # f° by Proposition 8.2.10 and the Maass lift Fy € S,(['z) of f is non-zero. Let e denote the adic cyclotomic character By Proposition 6.2.3, the standard T-function of Fy factors as
Ly (s, Fz, 1) = L(BC(f),s -2+k/2,w) L(BC(ƒ),s —3+k/2,w), hence the Galois representation pr, is reducible and has the form
Selmer groups 2 ee 163
We keep the notation from the previous section We now define the Selmer group relevant for our purposes For a field L and a Gr-module M (where we assume the action of Gz on M to be continuous) we put H'(L, M) to be a shorthand for the group Hin.(Gr,M) of cohomology classes of continuous cocycles Œy — M If L is a number field and p a finite place of L, set
Hi.,(Lp, M) := ker{H'(Ly, M) + H*(Ip, M)}.
Notation 10.2.1 In this and the next section, the letters V, V’, V”, V will denote Galois representations over an ¢-adic field E (as in section 10.1), and 7, T’, T”, T will be some O-lattices inside those representations We also set W := V/T, W’ := V'/T", etc and W, := W[A"] = {u EW | Aw = 0} S T/T, Wh := W'[A"], ete.
We begin by defining local Selmer groups For every finite prime p † £ we simply put
Hệ (Ly, V) := Hy (by, V), but for p | ý the definition is more subtle Let Bayi; denote Fontaine’s ring of -adic periods (cf [Fon82]) If p | 2 we set
For p | £, we call the D,-module V crystalline (or the Gr-module V crystalline at p) if dima, V = dima, H°(Lp, V @ Beis) When we refer to a Galois representation p: Gz, — GL(V) as being crystalline at p, we mean that V with the Gr-module structure defined by ứ is cristalline at p Also, for every p we define (Hy, W) to be the image of H}(y, V) under the natural map H'(L,,V) — H}(Ly,W) Using the fact that Gal(p : Ky) = Z has cohomological dimension 1, one easily sees that for p { £, H? (Ly, W) = Hj, (Ly, W) (cf [Bro05], Lemma VIII.15).
H}(L,W) := ker ¢ H1(L,W) => —————- rà) mel) HỆ (Lp, W) te
165 is called the (global) Selmer group of W.
The adjoint representation 2 1 ng gu gà ga 165
We keep the notation from the previous sections Let G be a group, R a commuta- tive ring with identity, M a finitely generated free R-module with an A-linear action of G given by a representation ứ : G + GLpr(M) We denote by ứY the contragredient of p One checks easily that after fixing an R-basis of Ä and interpreting the auto- morphisms p(g) as matrices with respect to that basis, we have p’(g) = (o(g)*)7}. Hence in particular if ứ is a character then pY % ứ~1 If M is of rank 2 over R, we have one more identification ứY & (det” op) @ ứ, which comes from the fact that for any 2 x 2 matrix A, (A*)~! = (det 4)~![; !]”” A[, 1] and the representations p and [i1] *sứâ[Ă !] are obviously isomorphic.
For any two representations ứ; : Ở —> GLr(M;), 7 = 1,2, the R-module
Homa(p1, ỉ2) is naturally a G-module with the action given by
(9 - Ó)(0) = p2(9)$(er(97")2), and there is an isomorphism of /2đ]-modules
We set ad p = Homa(p, ứ) = pí @ p.
By the discussion above we have pY @ p & (detTM' op) @p@p.
Let M, M’ and M” be finitely generated free R-modules with an action of G given by representations ứ, p’ and ứ”, respectively, and suppose that Ä is an R[/G|-module Py P ỉ extension of M” by M’, i.e., there is a short exact sequence of R[G]-modules
Let {ej 7.1 be an R-basis of Mƒ” For every j = 1, ,n choose e; € B-* (eh) The map sx : M" — M given by e; +> e; defines an A-linear section to 3, which is well-defined since M” is free (Note sx is not necessarily R[G]-linear.) Set ox: Ga Homn(/”, 8)
Lemma 10.3.1 The map X + @x defines a bijection between the set of extensions
Proof The proof is exactly the same as the proof of Proposition 4 in [Was97] TL]
Let # be a finite extension of Q¿ Let ứ : Gz - GLp(V'), p”: Gr — GLg(V”) be two Galois representations Choose O-lattices T’ C V’, T” C V”, and define W’, W”, etc as in Notation 10.2.1 Assume ỉ and p” are irreducible and short at p | Ê (for a definition of short, refer to [DFG04], section 1.1.2).
Set V = Homzg(p”,p') and T = Homao(/ỉ7„, py), and W = V/T Let p be a finite place of L Lemma 10.3.1 provides a natural bijection between H!(Lằ, W,) and Extprp,)(W,, Wy).
We now define the degree n local Selmer groups If p { £, set
Hj (Lp, Wn) := Hin(Lp, Wn), where W is as above.
If p | £, define Hj(Ly,W,) C H'(Ly, Wn) to be the subset consisting of those coho- mology classes which correspond to extensions
167 such that W, is in the essential image of the functor V defined in [DFG04], section
1.1.2) We will not need the precise definition of V It was shown in [DFG04] that
Hệ (Ty, Wa) is an O-submodule of Z1 (Dạ, W„) and that HỆ (Ly, W,,) is the preimage of H}(Ly,Wn+i) under the natural map H7(Ly,W,) > H}(Lp,Wnr+i) (ef Section 2.1, loc cit.).
Lemma 10.3.2 Let | £ be a place of L, 6: Gp — GLg(V) a Galois representation short atp, T C V an O[D,]-stable lattice and W := V/T If W,, fits into an exact sequence
0 — W¿ ơ W„ = W¿ >0 € Exto/p,(WZ, W2), then such an extension gives rise to an element of Hệ (Ly, Wn).
Proof See [DFG04], Section 1.1.2 Oo
Proposition 10.3.3 The natural isomorphism lim H (Ly, Wa) % (Uy, W) n induces a natural isomorphism lim H}(Lp, Wa) © H (Lạ, W). n
Proof See [DFG04], Proposition 2.2 oO
We now make specific choices for the representations for (V’, ứ) and (V”, p”) Let N® be a basis of eigenforms of S,([z) From now on we choose #7 to be a finite extension of Q¿ containing the fields Ep from Theorem 10.1.3 for all F € N* Let O be the valuation ring of with uniformizer À By Theorem 5.2.14, the eigenvalues of all T € Tà lie in O for all F € N* Let f € S,_1 (4, (=)) be a normalized eigenform with ƒ # f? and let Fy be the Maass lift of f Suppose there exists a non-CAP cuspidal eigenform Ƒ” € S,(I'z) with Hecke eigenvalues (for all T € TỆ,) congruent to those of Fy mod À Set pr x := ỉ/le„ We will consider the representations P7,K, Pr as having coefficients in the same ring O and their reductions as having coefficients in F := O/X From now on set 6 := pr, ỉ := pzK, 0” := 0g Se, and p := Homg(p",p’) Let V,V’, V” and V denote the corresponding representation spaces Assume pr is irreducible By Proposition 10.4.1 below there exists a lattice
T C V such that the induced (mod À) representation W; on which Œy acts via Dr: Ge —= GLp(T/AT) has the form
PrK OE and is not semisimple Here € denotes the mod @ cyclotomic character The map ở can be thought of as a non-trivial element of
Homp(p, đÊ,ỉ¿x) = (5; @8)’ đ7ỉ¿ = Py @ ỉ;y @ Ê' Sad PpK @ e,
There is an exact sequence
X: 0—>?P;,K —— Pp — PK @E—>0, with a section 8x : Dy % @€ — Py of F-vector spaces (not Gx-modules) coming from the direct sum decomposition (of vector spaces)
Hence by Lemma 10.3.1, the extension X gives rise to a non-zero element ox € H'(Gx, W,).
Theorem 10.3.4 Let £ > k be a prime Assume that py x 1s irreducible LetF" € &(Tz) be a non-CAP eigenform whose Hecke eigenvalues are congruent (mod
169 À) to those of Fy Let pr: Gx — GLp(V) be the Galois representation attached to F’ and assume it is irreducible There exists a Gx-stable O-lattice T CV such that the mod À representation Dp (where the reduction is taken with respect to XT) as ?sornorphic to
PreK OE where the cocycle represents a non-trivial class in the group H}(K, W).
Proof The existence of the lattice T such that J, is of the desired form (in par- ticular non-semisimple) follows from Proposition 10.4.1 which will be proved in the next section Lemma 10.3.1 implies that the cocycle ý gives rise to a non- trivial cohomology class [¢] € H!(K,W,) We first show that the natural map H’(K,W,) > H'(K,W) is injective Recall that the action of Gx on adj, x is given by œ + ỉ;x(g)owo7ỉy(g~”) Since this action does not change the deter- minant of u, and Ê # 2, there is no subspace of ad7ỉ;„ on which Gx acts via the character € Thus H°(K,W,) = 0 Note that we also must have H°(K,W) = 0. Indeed, let a € H°(K,W) Since every element of W is killed by a power of À, if œ # 0 there exists n such that À*œ = 0, but À*~!œ # 0 If this is the case however, then \”~la € H1(K,W,) = 0, which yields a contradiction Hence H°(K,W) = 0.
(which is exact since W is \-divisible) gives rise to the canonical injection and we conclude that we can regard [ở] as a non-trivial element of H!(K,W).
It remains to show that [¢] € H7(K,W) As adj, @€7 is unramified at all p { 22, it is clear that the local conditions at p { 2/ are satisfied We first deal with the prime p = +1) lying over (2) As W is unramified at p, we have H' (Ip, W) = H*(I?*, W), where J, ab is the Galois group of the maximal abelian subextension of K,/K, p Since
21, H'(1°,W) % H'((129)°29,W), where (73>)*TM* & lim F>, is the Galois group of the maximal tamely ramified extension of K,” We have
H`(P°, W) = Hom(lim Fặ,, W) = lim Hom(Eš,, W).
The group Hom(F3,,W) # 0 only if ý | (2" — 1) However, if | (2” — 1), then #{ (2"** —1) and Hom(F}.41,W) = 0 Hence lim Hom(Fš,, W7) = 0 Thus H*(I,,W) 0 and [ở] is unramified at 2.
It remains to check that [¢] satisfies the local conditions at p | 2 However, this is clear by Lemma 10.3.2 since by Theorem 10.1.3, pr is short at p L
Remark 10.3.5 The assumption that pz in Theorem 10.3.4 be irreducible is pre- sumably unnecessary In fact, it is expected that if an automorphic representation of G(A) is neither CAP nor endoscopic, then the associated (¢-adic) Galois repre- sentation is irreducible.
A non-semisimple reduction of a Galois representation 170 BIBLIOGRAPHY 0.0.0 .00 00 2 ee 174
In this section we do not adhere to the notation we have been using so far Let
E be a finite extension of Q; with valuation ring ©, uniformizer À and residue field F = O/A Let V be a 2n-dimensional E-vector space Here n is a positive integer Let G be a compact topological group and p: G — GLg(W) be a continuous irreducible representation By compactness of G there exists a G-stable O-lattice
AC V Then p gives rise to the representation ỉA : G — GLp(A/AA) Assume that for each G-stable O-lattice the representation p, stabilizes an n-dimensional E-subspace W of A/AA Write A/AA = W, @ W2 (as vector spaces), where W, is an n-dimensional F-subspace of A/AA complementary to Wị Then p, has the form [7 b|, with o, € GLp(W;), o2 € GLg(MW2) and f € Homp(W2, W1), and its semisimplification (whose isomorphism class is independent of A) has the form [7 4 ] We assume that ơi and o are irreducible and non-isomorphic.
This section is devoted to proving the following proposition.
Proposition 10.4.1 There exists a G-stable O-lattice A such that BA has the form ơi with ơy,ơa and f as above, but is not semi-simple.
We begin with the following lemma.
Lemma 10.4.2 Let p,,p2 : G — GL,(F) be two continuous, irreducible, non- isomorphic representations, and assume that ơ : G — GLan(E) given by p2(9) is semi-simple Here ƒ € Homp(W2,W1) and W; is the representation space of p;,
7 = 1,2 Let U denote the unipotent radical of the maximal parabolic subgroup P of Glo, stabilizing the flag 0 CW, CW, @WWa There exists u € U(F) such that
Proof Note that |” ằ.] is a semisimplification of ơ Since o itself is semisimple there exists h € GLa„(F) such that hơh”! = [?!;;] as any two semisimplifications of the same representation are isomorphic Assume h = [4 8] with A, B,C, D being n x n-matrices Then hah! = [” „„] implies that pi A Blip f * * ứ2'| |Œ D p2 pz Cpy' * C hence C € Homg(W), Wo) = 0 as pi, 2 are irreducible and non-isomorphic Thus h € P(F) On the other hand, [4° p-1|h € U(F) and we have
So, replacing h with [47 p- | h we can assume h € U(F) and diagonalizeso O
Proof of Proposition 10.4.1 We follow [Rib76], Proposition 2.1 Choose a G-stable lattice A C V, together with an O-basis B of this lattice Then ứ may be viewed as a map G — GLo,(O) If M € GLan(E) is a matrix such that Mo(G)M-! c GLo,(O) then MA C V is another G-stable O-lattice with an O-basis MB.
Choose a G-stable O-lattice A and an O-basis B of it After reducing p modulo
A, the basis B gives rise to an F-basis with respect to which the reduction p, of p has the form [* 5,], with ứ;(g) € GLe(W;) for every g € G and ứ; = ơ; for
7 = 1,2 Furthermore we will assume that for every such lattice A, the representation
Pa = [” oo | is semi-simple and derive a contradiction from this.
Let Tạ denote the n x n-identity matrix Let Mp be the 2n x 2n-identity matrix.
Inductively we will define a convergent sequence of matrices M; = [’ nae 1 € (O)", such that M,p(G)M;"* consists of elements of GL2,(O) whose lower-left n x ? n-block is of the form AS;, where 5; € (Ó} and whose upper-right n x n-block n is of the form Afổ;, where S; € (O)" This will force ứ to be reducible because n (im; M;) p(G) (limi M7") consists of matrices whose upper-right n x n-blocks are 0, but ứ is irreducible by assumption.
We now carry out the induction Our induction hypothesis is as follows:
Mip(G)M; À1 AIn consists of elements of GLa„(Œ) whose lower-left n x n-block is of the form \'t1S; with S; € (O)? If i = 0, the assumption is clearly satisfied, hence we assume it is true for 7 = 7 and prove it fori = 7 +1 By the inductive assumption, j -j
AT, AIn is a representation whose mod À reduction is of the form |“ mi | with p, & ứ; for j = 1,2 Thus ứĂ and ứ¿ are irreducible and non-isomorphic Applying Lemma
10.4.2 with o equal to the (mod A) reduction ỉ of p’ we conclude that there exists
T € (F)? such that | H (9) |” H is diagonal Let T denote any lift of T to
O Then |” 7 | ứ(đ) |” ry consists of elements of GL2,(O) whose upper-right n x n-blocks are of the form À5; with 6; € (O)f and whose lower-left n x n-blocks are of the form À/*1ỉ', with $’; € (0) Hence
Int M; | p(9) À1, hạ An consists of elements of GLa„() whose upper-right n x n-blocks are of the form
778541 with S;41 € (O) Here Int (g)(A) := gAgTM! for any matrix A and any invertible matrix g So, we set
[BCdS?03] D Bump, J W Cogdell, E de Shalit, D Gaitsgory, E Kowalski, and S S Kudla.
An introduction to the Langlands program Birkhauser Boston Inc., Boston, MA, 2003. Lectures presented at the Hebrew University of Jerusalem, Jerusalem, March 12-16,
2001, Edited by Joseph Bernstein and Stephen Gelbart.
Don Blasius and Jonathan D Rogawski Zeta functions of Shimura varieties In Motives (Seattle, WA, 1991), volume 55 of Proc Sympos Pure Math., pages 525-571 Amer. Math Soc., Providence, RI, 1994.
Jim Brown Saito-Kurokawa lifts, L-values for GLo, and congruences between Siegel modular forms Thesis University of Michigan, Ann Arbor, 2005.
Daniel Bump Automorphic forms and representations, volume 55 of Cambridge Studies in Advanced Mathematics Cambridge University Press, Cambridge, 1997.
Henri Darmon, Fred Diamond, and Richard Taylor Fermat’s last theorem In Ellip- tic curves, modular forms & Fermat’s last theorem (Hong Kong, 1993), pages 2-140. Internat Press, Cambridge, MA, 1997.
Fred Diamond, Matthias Flach, and Li Guo The Tamagawa number conjecture of adjoint motives of modular forms Ann Sci Ecole Norm Sup (4), 37, 2004.
Fred Diamond and John Im Modular forms and modular curves In Seminar on Fermat’s Last Theorem (Toronto, ON, 1993-1994), volume 17 of CMS Conf Proc., pages 39-133 Amer Math Soc., Providence, RI, 1995.
Bart de Smit and Hendrik W Lenstra, Jr Explicit construction of universal deforma- tion rings In Modular forms and Fermat’s last theorem (Boston, MA, 1995), pages 313-326 Springer, New York, 1997.
David Eisenbud Commutative algebra, volume 150 of Graduate Texts in Mathematics. Springer-Verlag, New York, 1995 With a view toward algebraic geometry.
Jean-Marc Fontaine Sur certains types de représentations p-adiques du groupe de Galois d’un corps local; construction d’un anneau de Barsotti-Tate Ann of Math (2), 115(3):529-577, 1982.
E Freitag Siegelsche Modulfunktionen, volume 254 of Grundlehren der Mathematis- chen Wissenschaften [Fundamental Principles of Mathematical Sciences] Springer- Verlag, Berlin, 1983.
Stephen S Gelbart Automorphic forms on adéle groups Princeton University Press, Princeton, N.J., 1975 Annals of Mathematics Studies, No 83.
Stephen Gelbart Three lectures on the modularity of ứp; and the Langlands reci- procity conjecture In Modular forms and Fermat’s last theorem (Boston, MA, 1995),pages 155-207 Springer, New York, 1997.
V A Gritsenko The zeta function of degree six for Hermitian modular forms of genus
2 Zap Nauchn Sem Leningrad Otdel Mat Inst Steklov (LOMI), 154(Anal Teor. Chisel i Teor Punktsii 7):46-66, 176-177, 1986.
V A Gritsenko The Maass space for SU(2,2) The Hecke ring, and zeta functions. Trudy Mat Inst Steklov., 183:68-78, 223-225, 1990 Translated in Proc Steklov Inst. Math 1991, no 4, 75-86, Galois theory, rings, algebraic groups and their applications (Russian).
V A Gritsenko Parabolic extensions of the Hecke ring of the general linear group IT. Zap Nauchn Sem Leningrad Otdel Mat Inst Steklov (LOMI), 183(Modul Funktsii i Kvadrat Formy 1):56-76, 165, 167, 1990.
Haruzo Hida Galois representations and the theory of p-adic Hecke algebras Suứaku, 39(2):124-139, 1987 Sugaku Expositions 2 (1989), no 1, 75-102.
Haruzo Hida Elementary theory of L-functions and Eisenstein series, volume 26 of London Mathematical Society Student Texts Cambridge University Press, Cambridge, 1993.
Haruzo Hida Modular forms and Galois cohomology, volume 69 of Cambridge Studies in Advanced Mathematics Cambridge University Press, Cambridge, 2000.
Haruzo Hida p-adic automorphic forms on Shimura varieties Springer Monographs in Mathematics Springer-Verlag, New York, 2004.
Tatsuo Hina and Takashi Sugano On the local Hecke series of some classical groups over p-adic fields J Math Soc Japan, 35(1):133-152, 1983.
Henryk Iwaniec Topics in classical automorphic forms, volume 17 of Graduate Studies in Mathematics American Mathematical Society, Providence, RI, 1997.
H Jacquet and J A Shalika On Euler products and the classification of automorphic forms II Amer J Math., 103(4):777-815, 1981.
H Jacquet and J A Shalika On Euler products and the classification of automorphic forms II Amer J Math., 103(4):777-815, 1981.
Henry H Kim Automorphic L-functions In Lectures on automorphic L-functions, volume 20 of Fields Inst Monogr., pages 97-201 Amer Math Soc., Providence, RI, 2004.
Takuya Kon-No The residual spectrum of U(2,2) Trans Amer Math Soc., 350(4):1285-1358, 1998.
A W Knapp Introduction to the Langlands program In Representation theory and automorphic forms (Edinburgh, 1996), volume 61 of Proc Sympos Pure Math., pages 245-302 Amer Math Soc., Providence, RI, 1997.
Hisashi Kojima An arithmetic of Hermitian modular forms of degree two Invent. Math., 69(2):217-227, 1982.
Aloys Krieg The MaaB spaces on the Hermitian half-space of degree 2 Math Ann., 289(4):663-681, 1991.
Robert P Langlands On the functional equations satisfied by Eisenstein series. Springer-Verlag, Berlin, 1976 Lecture Notes in Mathematics, Vol 544.
Hideyuki Matsumura Commutative ring theory, volume 8 of Cambridge Studies inAdvanced Mathematics Cambridge University Press, Cambridge, second edition, 1989.Translated from the Japanese by M Reid.