Home Period relations for cusp forms of GSp4
Article
Licensed
Unlicensed Requires Authentication

Period relations for cusp forms of GSp4

  • Harald Grobner ORCID logo EMAIL logo and Ronnie Sebastian
Published/Copyright: August 15, 2017

Abstract

Let F be a totally real number field and let π be a cuspidal automorphic representation of GSp4(𝔸F), which contributes irreducibly to coherent cohomology. If π has a Bessel model, we may attach a period p(π) to this datum. In the present paper, which is Part I in a series of two, we establish a relation of these Bessel periods p(π) and all of their twists p(πξ) under arbitrary algebraic Hecke characters ξ. In the appendix, we show that (𝔤,K)-cohomological cusp forms of GSp4(𝔸F) all qualify to be of the above type – providing a large source of examples. We expect that these period relations for GSp4(𝔸F) will allow a conceptual, fine treatment of rationality relations of special values of the spin L-function, which we hope to report on in Part II of this paper.


Communicated by Freydoon Shahidi


Funding statement: The first author is supported by the Austrian Science Fund (FWF), START-prize Y966-N35 and stand-alone-research project P 25974-N25. The second author was partly supported by a DST INSPIRE grant.

A General aspects in the automorphic theory of the cohomology of Shimura varieties

A.1 Two relative Lie algebra cohomology theories related to Shimura varieties

In this appendix, we let (G,X) be the datum defining a Shimura variety in the sense made precise in [12, Section 1.1]. For the sake of completeness, we recall that this means that G is a connected reductive linear algebraic group over and X is a G()-conjugacy class of homomorphisms h:R/(GL1)G× such that the following conditions hold:

  1. The Hodge structure on the Lie algebra 𝔤 of G given by Adh is of type (0,0)+(1,-1)+(-1,1).

  2. The automorphism Ad(h(i)) induces a Cartan involution on G𝗌𝗌(), with G𝗌𝗌 being the derived group of G. The -group G𝗌𝗌× has no anisotropic factors over .

  3. The weight map hw:GL1×G×, where w:GL1×R/(GL1) is the canonical co-norm map, is defined over .

  4. For a maximal -split torus ZZG, the quotient ZG()/Z() is compact.

With these assumptions, X is the finite disjoint union of Hermitian symmetric spaces of the form G𝗌𝗌()/K𝗌𝗌, where K𝗌𝗌 is a maximal connected compact subgroup of G𝗌𝗌(). We let K be the centralizer of a fixed point hX in G(). It contains the product of K𝗌𝗌 and ZG(). Let 𝔨 be the Lie algebra of K. It operates by the adjoint action on 𝔤, and we obtain a 𝔨-invariant decomposition

(A.1)𝔤=𝔭+𝔨𝔭-.

Here, 𝔭- (resp. 𝔭+) is the holomorphic (resp. anti-holomorphic) tangent space of X at h. We let

𝔭h:=𝔨𝔭+.

This is a parabolic subalgebra of 𝔤 with Levi subalgebra 𝔨 and nilpotent, even abelian, radical 𝔭+. Observe that 𝔭h lies somewhat “skew” to the real structure of 𝔤=𝔤i𝔤 as 𝔭h𝔤=𝔨.

For us, a 𝔤-module V, which is also a representation of K, is called a (𝔤,K)-module if it is a (𝔤𝗌𝗌,K𝗌𝗌)-module in the sense of Borel–Wallach [4, Section 0.2] by restriction. Mainly to set notation and for the sake of precision, we will now rapidly recall the definition of two relative Lie algebra cohomology theories.

The relative Lie algebra cohomology Hq(𝔤,𝔨,V) of V was defined in [4, Section I.1.2]. In [4, Section I.5.1], also the (𝔤,K)-cohomology of V was defined. It is the cohomology Hq(𝔤,K,V) of the complex

Cq(𝔤,K,V):=HomK(Λq(𝔤/𝔨),V)HomK(Λq(𝔭+𝔭-),V),
df(X0,,Xq):=i=0q(-1)iXif(X0,,X^i,,Xq)+i<j(-1)i+jf([Xi,Xj],X0,,X^i,,X^j,,Xq).

If K is connected, then Hq(𝔤,𝔨,V)=Hq(𝔤,K,V). We recall that an irreducible representation Vλ of the real Lie group G() on a finite-dimensional complex vector space is called algebraic if the (extended) action of the complex Lie group G() on Vλ is a representation of the linear algebraic group G× over . Finally, we say that a (𝔤,K)-module V is (𝔤,K)-cohomological if there is an irreducible finite-dimensional algebraic G()-module Vλ such that Hq(𝔤,K,VVλ)0 for some degree q.

The (𝔭h,K)-cohomology of a (𝔤,K)-module V is the cohomology of the complex

Cq(𝔭h,K,V):=HomK(Λq(𝔭h/𝔨),V)HomK(Λq𝔭+,V),

with df defined as above. Following [13], we say that a (𝔤,K)-module V is (𝔭h,K)-cohomological if there is an irreducible finite-dimensional K-module Vτ such that the space Hq(𝔭h,K,VVτ)0 for some degree q.

A.2 A general result on the relation of (𝔤,K)-cohomology and (𝔭h,K)-cohomology

We denote by 𝒞G (resp. 𝒞K) the collection of all equivalence classes of irreducible algebraic representations of G() (resp. irreducible finite-dimensional representations of K). We assume to have fixed a Cartan subalgebra 𝔱 of 𝔨 and a set of positive roots Δ+(𝔨,𝔱). Because of (A.1), 𝔱 is a Cartan subalgebra of 𝔤, too, and we assume that Δ+(𝔤,𝔱) is a choice of positive roots for 𝔤 with respect to 𝔱, extending the given choice Δ+(𝔨,𝔱) for 𝔨. We assume that 𝔭h is a standard parabolic subalgebra of 𝔤 with respect to Δ+(𝔤,𝔱), i.e., all roots in 𝔭+ are positive (which also explains the notation).

Clearly, 𝔭¯h:=𝔨𝔭- is the parabolic subalgebra of 𝔤, which is opposite to 𝔭h. If Vλ𝒞G, then Vλ is determined by its highest weight λ with respect to Δ+(𝔤,𝔱). Similarly, if Vτ𝒞K, then Vτ is determined by its highest weight τ with respect to Δ+(𝔨,𝔱). Let

ρ:=12αΔ+(𝔤,𝔱)α(resp. ρc:=12αΔ+(𝔨,𝔱)α)

be the half-sum of roots in Δ+(𝔤,𝔱) (resp. Δ+(𝔨,𝔱)), and let W (resp. W𝔨) be the Weyl group of 𝔤 (resp. 𝔨) with respect to 𝔱. Then the infinitesimal characters χVλ (resp. χVτ) of Vλ (resp. Vτ) are determined by λ+ρ (resp. τ+ρc) up to the action of W (resp. W𝔨). We will use the notation χVλ=χλ+ρ and χVτ=χτ+ρc. Recall that there is the obvious surjection

ξ:Hom(Z(𝔨),)Hom(Z(𝔤),),

cf. [13, p. 31], mapping χΛ onto ξ(χΛ)=χΛ+ρn. Here, Z(𝔨) (resp. Z(𝔤)) denotes the center of the universal enveloping algebra of 𝔨 (resp. 𝔤), and ρn=ρ-ρc is the half-sum of non-compact roots in Δ+(𝔤,𝔱) (i.e., the roots appearing in 𝔭+). The following is the main result of our appendix.

Theorem A.1.

Let V be an irreducible unitary (g,K)-module and let VλCG. If V is (g,K)-cohomological with respect to Vλ in degree q, then V is (ph,K)-cohomological in some degree aq.

Proof.

Let V and Vλ be as in the statement of the proposition and assume that V is (𝔤,K)-cohomological with respect to Vλ. Hence,

HomK(Λq(𝔭+𝔭-),VVλ)0

for some degree q. As

HomK(Λq(𝔭+𝔭-),VVλ)a+b=qHomK(Λa𝔭+Λb𝔭-,VVλ),

there are 0a,bq such that a+b=q and

HomK(Λa𝔭+Λb𝔭-,VVλ)HomK(Λa𝔭+,V(VλΛb𝔭-*))0.

Observe that there is an isomorphism of K-representations VλΛb𝔭-*Hb(𝔭-,Vλ). We may hence use Kostant’s description of the K-module Hb(𝔭-,Vλ): To this end, we identify 𝔭- as the nilpotent radical of the parabolic subalgebra 𝔭¯h𝔤 opposite to 𝔭h. Let W𝔭h be the set of Weyl group elements w for which w-1(α)Δ+(𝔤,𝔱) for all αΔ+(𝔨,𝔱). Going over to the opposite ordering of roots, we obtain the set of Kostant representatives for 𝔭¯h as W𝔭¯h=wGW𝔭hwG. Here, wG denotes the longest element of W. Therefore, [18, Theorem 5.14], implies that there is an isomorphism of K-modules

Hb(𝔭-,Vλ)wW𝔭¯h(w)=bVwc(w(wG(λ)+ρ¯)-ρ¯),

where wc is the longest element in the Weyl group W𝔨 of K, ρ¯ is the half-sum of positive roots with respect to the opposite ordering in Δ(𝔤,𝔱) (whence equal to ρ¯=wG(ρ)=-ρ) and Vτ denotes the irreducible K-representation of highest weight τ with respect to Δ+(𝔨,𝔱). A simple calculation using [4, Section V.1.4] hence implies that as K-modules

(A.2)Hb(𝔭-,Vλ)wW𝔭¯h(w)=dim(𝔭-)-bVw(λ+ρ)+wc(ρ).

Abbreviate b:=dim(𝔭-)-b and τw:=w(λ+ρ)+wc(ρ), and let Vτw be the corresponding irreducible K-representation appearing in (A.2). Hence, by what we have seen above, we have proved that there are 0a,bq such that a+b=q and

0HomK(Λa𝔭+,V(VλΛb𝔭-*))wW𝔭¯h(w)=bHomK(Λa𝔭+,VVτw).

Hence, there is a wW𝔭¯h of length (w)=b such that

HomK(Λa𝔭+,VVτw)0.

Fix such a Kostant representative wW𝔭¯h. The infinitesimal character of the contragredient Vτw𝗏 is given by

χVτw𝗏=χ-wc(τw)+ρc=χ-wcw(λ+ρ)-ρ+ρc=χ-wcw(λ+ρ)-ρn,

and hence mapped by the surjection ξ onto the infinitesimal character of the contragredient of the algebraic G()-representation Vλ:

(A.3)ξ(χVτw𝗏)=χ(-wcw(λ+ρ)-ρn)+ρn=χ-wcw(λ+ρ)=χ-λ-ρ=χVλ𝗏.

Let C𝔤 be the Casimir operator in Z(𝔤). Then, as by assumption V is an irreducible unitary (𝔤,K)-module, which is (𝔤,K)-cohomological with respect to Vλ, the infinitesimal character χV of V and the infinitesimal character χVλ of Vλ agree on C𝔤 as a consequence of [4, Proposition II.3.1]. In particular, [3, Lemma 1.3] and (A.3) imply that

χV(C𝔤)=ξ(χVτw𝗏)(C𝔤).

The analogue of Kuga’s formula for (𝔭h,K)-cohomology, established as [20, Theorem 4.1] (see also [13, Proposition 4.4.3]), hence implies verbatim as in the proof of [4, Proposition II.3.1.(b)] that

HomK(Λa𝔭+,VVτw)=Ha(𝔭h,K,VVτw).

So, by our above discussion, we finally obtain

Ha(𝔭h,K,VVτw)0.

Thus, V is (𝔭h,K)-cohomological in degree aq with respect to some irreducible K-summand Vτw of VλΛb𝔭-*. ∎

We conclude this appendix by the following corollary.

Corollary A.2.

Let V be an irreducible unitary (g,K)-module and let VλCG. The degrees q, where V is (g,K)-cohomological with respect to Vλ, are all of the same parity.

Proof.

By using the Künneth rule, [4, Section I.1.3] and [25, Theorem 5.5], it is enough to show this for the trivial (𝔤,K)-module V=𝟏 and the trivial coefficient system Vλ= of G(). It is clear that V=𝟏 has non-trivial (𝔤,K)-cohomology with respect to Vλ= in degree q=0. Hence, we have to show that all degrees q, where Hq(𝔤,K,𝟏)0, are even. Let q be any such degree. Then, as we have seen in the proof of Proposition A.1, there are a, b such that q=a+b and

HomK(Λa𝔭+,Λb𝔭-*)0.

In other words, the K-representations Λa𝔭+ and Λb𝔭-*Λb𝔭+ share an irreducible K-type. For any r, we have

Λr𝔭+i1,,irj=1r𝔤αij,

where 𝔤αij is the one-dimensional root eigenspace of 𝔤 of the non-compact, positive root αij. Hence, for Λa𝔭+ and Λb𝔭+ to have an irreducible K-type in common, we have to have a=b, whence, q=2a is even as predicted. ∎

Acknowledgements

We thank A. Raghuram for suggesting to work on this project. Further, we thank him for useful discussions. We are also grateful to the anonymous referee for her/his careful reading and numerous, very helpful suggestions, which improved the presentation of the present article.

References

[1] M. Asgari and F. Shahidi, Generic transfer from GSp(4) to GL(4), Compos. Math. 142 (2006), no. 3, 541–550. 10.1112/S0010437X06001904Search in Google Scholar

[2] D. Blasius, Period relations and critical values of L-functions, Pacific J. Math. (1997), Special Issue, 53–83, 10.2140/pjm.1997.181.53Search in Google Scholar

[3] A. Borel and W. Casselman, L2-cohomology of locally symmetric manifolds of finite volume, Duke Math. J. 50 (1983), no. 3, 625–647. 10.1215/S0012-7094-83-05029-9Search in Google Scholar

[4] A. Borel and N. Wallach, Continuous Cohomology, Discrete Subgroups, and Representations of Reductive Groups, 2nd ed., Math. Surveys Monogr. 67, American Mathematical Society, Providence, 2000. 10.1090/surv/067Search in Google Scholar

[5] P. Deligne, Travaux de Shimura, Séminare Bourbaki 13 (1970/71), no. 389, 123–165. 10.1007/BFb0058700Search in Google Scholar

[6] P. Deligne, Valeurs de fonctions L et périodes d’intégrales, Automorphic Forms, Representations and L-Functions (Corvallis 1977), Proc. Sympos. Pure Math. 33 Part 2, American Mathematical Society, Providence (1979), 313–346. 10.1090/pspum/033.2/546622Search in Google Scholar

[7] M. Furusawa, On L-functions for GSp(4)×GL(2) and their special values, J. Reine Angew. Math. 438 (1993), 187–218. 10.1515/crll.1993.438.187Search in Google Scholar

[8] R. Godement, Notes on Jacquet–Langlands’ theory, Matematika 18 (1974), no. 3, 3–45. Search in Google Scholar

[9] H. Grobner and M. Harris, Whittaker periods, motivic periods, and special values of tensor product L-functions, J. Inst. Math. Jussieu 15 (2016), no. 4, 711–769. 10.1017/S1474748014000462Search in Google Scholar

[10] H. Grobner and A. Raghuram, On the arithmetic of Shalika models and the critical values of L-functions for GL2n, Amer. J. Math. 136 (2014), no. 3, 675–728. 10.1353/ajm.2014.0021Search in Google Scholar

[11] G. Harder, General aspects in the theory of modular symbols, Séminaire de Théorie des Nombres (Paris 1981/82), Progr. Math. 38, Springer, Boston (1983), 73–88. Search in Google Scholar

[12] M. Harris, Arithmetic vector bundles and automorphic forms on Shimura varieties. I, Invent. Math. 82 (1985), no. 1, 151–189. 10.1007/BF01394784Search in Google Scholar

[13] M. Harris, Automorphic forms of ¯-cohomology type as coherent cohomology classes, J. Differential Geom. 32 (1990), no. 1, 1–63. 10.4310/jdg/1214445036Search in Google Scholar

[14] M. Harris, L-functions and periods of polarized regular motives, J. Reine Angew. Math. 483 (1997), 75–161. 10.1515/crll.1997.483.75Search in Google Scholar

[15] M. Harris, Occult period invariants and critical values of the degree four L-function of GSp(4), Contributions to Automorphic Forms, Geometry, and Number Theory, Johns Hopkins University Press, Baltimore (2004), 331–354. Search in Google Scholar

[16] T. Ikeda and S. Yamana, On the lifting of Hilbert cusp forms to Hilber–Siegel cusp forms, preprint (2015), https://arxiv.org/abs/1512.08878. Search in Google Scholar

[17] M. Kontsevich and D. Zagier, Periods, Mathematics Unlimited—2001 and Beyond, Springer, Berlin (2001), 771–808. 10.1007/978-3-642-56478-9_39Search in Google Scholar

[18] B. Kostant, Lie algebra cohomology and the generalized Borel–Weil theorem, Ann. of Math. (2) 74 (1961), 329–387. 10.2307/1970237Search in Google Scholar

[19] J. Mahnkopf, Cohomology of arithmetic groups, parabolic subgroups and the special values of L-functions on GLn, J. Inst. Math. Jussieu 4 (2005), no. 4, 553–637. 10.1017/S1474748005000186Search in Google Scholar

[20] K. Okamoto and H. Ozeki, On square-integrable ¯-cohomology spaces attached to hermitian symmetric spaces, Osaka J. Math. 4 (1967), 95–110. Search in Google Scholar

[21] A. A. Panchishkin, Motives over totally real fields and p-adic L-functions, Ann. Inst. Fourier (Grenoble) 44 (1994), no. 4, 989–1023. 10.5802/aif.1424Search in Google Scholar

[22] I. I. Piatetski-Shapiro, L-functions for GSp4, Pacific J. Math. (1997), no. Special Issue, 259–275. 10.2140/pjm.1997.181.259Search in Google Scholar

[23] D. Prasad and R. Takloo-Bighash, Bessel models for GSp(4), J. Reine Angew. Math. 655 (2011), 189–243. 10.1515/crelle.2011.045Search in Google Scholar

[24] A. Raghuram and F. Shahidi, On certain period relations for cusp forms on GLn, Int. Math. Res. Not. IMRN 2008 (2008), Article ID rnn 077. Search in Google Scholar

[25] D. A. Vogan, Jr. and G. J. Zuckerman, Unitary representations with nonzero cohomology, Compos. Math. 53 (1984), no. 1, 51–90. Search in Google Scholar

[26] J.-L. Waldspurger, Quelques propriétés arithmétiques de certaines formes automorphes sur GL(2), Compos. Math. 54 (1985), no. 2, 121–171. Search in Google Scholar

[27] A. Weil, Basic Number Theory, Grundlehren Math. Wiss. 144, Springer, New York, 1967. 10.1007/978-3-662-00046-5Search in Google Scholar

Received: 2017-1-10
Revised: 2017-7-15
Published Online: 2017-8-15
Published in Print: 2018-5-1

© 2018 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 14.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/forum-2017-0005/html
Scroll to top button