This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Whittaker coefficients of geometric Eisenstein series

Jeremy Taylor Department of Mathematics
University of California, Berkeley
Berkeley, CA 94720-3840
jeretaylor@berkeley.edu
Abstract.

Geometric Langlands predicts an isomorphism between Whittaker coefficients of Eisenstein series and functions on the moduli space of Nˇ\check{N}-local systems. We prove this formula by interpreting Whittaker coefficients of Eisenstein series as factorization homology and then invoking Beilinson and Drinfeld’s formula for chiral homology of a chiral enveloping algebra.

1. Introduction

1.1. Notation and conventions

Let GG be a simply connected complex reductive group with Langlands dual group Gˇ\check{G} defined over k=𝐂k=\mathbf{C}. Choose a maximal torus TT and a Borel subgroup BB with unipotent radical NN. Let ρ\rho be half the sum of the positive coroots. Let XX be a smooth projective complex genus gg curve. Choose a square root of the canonical bundle on XX and form the anticanonical TT-bundle ωρ\omega^{-\rho}. All categories and functors are derived.

Let σ\sigma be a Tˇ\check{T}-local system on XX, and let LocNˇσ=LocBˇ×LocTˇσ\operatorname{Loc}_{\check{N}}^{\sigma}=\operatorname{Loc}_{\check{B}}\times_{\operatorname{Loc}_{\check{T}}}\sigma be the derived moduli stack of Bˇ\check{B}-local systems on XX whose underlying Tˇ\check{T}-local system is identified with σ\sigma, see (1.3). A Tˇ\check{T}-local system is called regular if for every coroot the associated rank 1 local system is nontrivial. If σ\sigma is regular then LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} is a classical affine scheme isomorphic to a vector space.

Let KK be the Hecke σ\sigma-eigensheaf on BunT\operatorname{Bun}_{T} whose stalk at ωρ\omega^{-\rho} twisted by a negative coweight valued divisor λ¯x¯=λixi\underline{\lambda}\cdot\underline{x}=\sum\lambda_{i}x_{i} is

(1.1) Kωρ(λ¯x¯)=(σxiλi)[dT+dλ].K_{\omega^{-\rho}(-\underline{\lambda}\cdot\underline{x})}=\left(\bigotimes\sigma^{-\lambda_{i}}_{x_{i}}\right)[d_{T}+d^{\lambda}].

Above, σxλ\sigma^{\lambda}_{x} means the fiber at xx of the rank 1 local system obtained from σ\sigma using λ\lambda. Here dT=dimBunTd_{T}=\dim\operatorname{Bun}_{T} and dλ=2ρˇ,4(g1)ρ+λd^{\lambda}=\langle 2\check{\rho},4(g-1)\rho+\lambda\rangle is the shift appearing in section 6.4.8 of [Gai13].

Remark.

Let KShvNilp(BunT)K^{\prime}\in\operatorname{Shv}_{\operatorname{Nilp}}(\operatorname{Bun}_{T}) correspond under class field theory to the skyscraper sheaf kσQCoh(LocTˇ)k_{\sigma}\in\operatorname{QCoh}(\operatorname{Loc}_{\check{T}}). The Hecke eigensheaf condition determines KK^{\prime} up to tensoring by a line. Whittaker normalization says that global sections of kσk_{\sigma} equals the costalk at the trivial TT-bundle of K[dT]K^{\prime}[d_{T}]. Thus KK is only noncanonically isomorphic to a shift of KK^{\prime}. On the degree λ2(g1)ρ-\lambda-2(g-1)\rho connected component BunTλ\operatorname{Bun}_{T}^{\lambda}, there is a canonical identification K=ωρK[dλ]K=\omega^{-\rho}K^{\prime}[d^{\lambda}]. We translated KK^{\prime} by ωρ\omega^{-\rho}, which has the effect of tensoring it by a certain line, see section 4.1 of [Gai10].

The Whittaker or Poincaré series sheaf Whit=r!χDexp=r!(χ)exp[2]\operatorname{Whit}=r_{!}\chi^{*}D\exp=r_{!}(-\chi)^{*}\exp[-2] on BunG\operatorname{Bun}_{G} is the pullback then pushforward of the exponential sheaf along

𝐀1𝜒BunNωρ𝑟BunG,\mathbf{A}^{1}\xleftarrow{\chi}\operatorname{Bun}_{N^{-}}^{\omega^{-\rho}}\xrightarrow{r}\operatorname{Bun}_{G},

see 5.4.1 of [FR22]. The function χ\chi is defined in for example [FGV01]. The character sheaf exp\exp on 𝐀1\mathbf{A}^{1} is normalized so that its costalks are in degree zero. Up to a shift, its Verdier dual DexpD\exp is the inverse character sheaf. Both exp\exp and DexpD\exp corepresent shifted vanishing cycles for conic sheaves on 𝐀1\mathbf{A}^{1}, so the distinction is not so important. In the Betti setting we do not have the exponential D-module. Because χ\chi is 𝐂×\mathbf{C}^{\times}-equivariant for the 2ρ2\rho action on BunNωρ\operatorname{Bun}_{N^{-}}^{\omega^{-\rho}} and the weight 2 action on 𝐀1\mathbf{A}^{1}, the sheaf defined in 2.5.2 of [NY19] serves as a substitute. The Whittaker sheaf does not have nilpotent singular support.

The automorphic and spectral Eisenstein series functors, Eis!=p!q\operatorname{Eis}_{!}=p_{!}q^{*} and Eˇis=pˇIndCohqˇIndCoh\operatorname{\check{E}is}=\check{p}^{\operatorname{IndCoh}}_{*}\check{q}^{\operatorname{IndCoh}*}, are defined by pullback then pushforward along

BunT𝑞BunB𝑝BunGandLocTˇqˇLocBˇpˇLocGˇ.\operatorname{Bun}_{T}\xleftarrow{q}\operatorname{Bun}_{B}\xrightarrow{p}\operatorname{Bun}_{G}\quad\text{and}\quad\operatorname{Loc}_{\check{T}}\xleftarrow{\check{q}}\operatorname{Loc}_{\check{B}}\xrightarrow{\check{p}}\operatorname{Loc}_{\check{G}}.

All of the above functors are left adjoints, in particular pˇIndCoh\check{p}^{\operatorname{IndCoh}}_{*} is a left adjoint because pˇ\check{p} is proper. In (1.2), the functor Eis!\operatorname{Eis}_{!} is modified according to section 4.1 of [Gai10] or section 6.4.8 [Gai13]. This matches the translation by ωρ\omega^{-\rho} and shift by dλd^{\lambda} built into our definition of KK in (1.1).

1.2. Main theorem statement

The geometric Langlands conjecture is supposed to be compatible with parabolic induction. Moreover the Whittaker functional is expected to correspond under Langlands to global sections on LocGˇ\operatorname{Loc}_{\check{G}}, up to a shift by dG=dimBunGd_{G}=\dim\operatorname{Bun}_{G}. Thus commutativity of conjectural diagram

(1.2) ShvNilp(BunT){\operatorname{Shv}_{\operatorname{Nilp}}(\operatorname{Bun}_{T})}{\simeq}QCoh(LocTˇ){\operatorname{QCoh}(\operatorname{Loc}_{\check{T}})}ShvNilp(BunG){\operatorname{Shv}_{\operatorname{Nilp}}(\operatorname{Bun}_{G})}{\simeq}IndCohNilp(LocGˇ){\operatorname{IndCoh}_{\operatorname{Nilp}}(\operatorname{Loc}_{\check{G}})}Vect{\operatorname{Vect}}Eis!((ωρ)[dλ])\scriptstyle{\operatorname{Eis}_{!}((\omega^{-\rho}-){[d^{\lambda}]})}Eˇis()\scriptstyle{\operatorname{\check{E}is}(-)}Hom(Whit,)[dG]\scriptstyle{\operatorname{Hom}(\operatorname{Whit}{,}-){[d_{G}]}}ΓIndCoh()\scriptstyle{\Gamma^{\operatorname{IndCoh}}(-)}

applied to the skyscraper kσk_{\sigma}, predicts the following isomorphism.

Main theorem.

Let σ\sigma be a Tˇ\check{T}-local system on XX and let KK be the Hecke eigensheaf on BunT\operatorname{Bun}_{T} defined in (1.1). Whittaker coefficients of Eisenstein series equals functions on moduli space of Nˇ\check{N}-local systems:

Hom(Whit,Eis!K)[dG]=𝒪(LocNˇσ).\operatorname{Hom}(\operatorname{Whit},\operatorname{Eis}_{!}K)[d_{G}]=\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma}).

The proof uses a combination of [Ras21] and [BG08] to relate twisted cohomology of the Zastava space to the formal completion of LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma}.

Both sides of the main theorem are coweight graded vector spaces. On the automorphic side, let KλK^{\lambda} be the restriction to the degree λ2(g1)ρ-\lambda-2(g-1)\rho connected component BunTλ\operatorname{Bun}_{T}^{\lambda}. On the spectral side, the adjoint Tˇ\check{T}-action on Bˇ\check{B} induces an action on LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma}.

Our results apply for all three versions of geometric Langlands: de Rham, restricted, and Betti. On the automorphic side Eis!K\operatorname{Eis}_{!}K is a constructible sheaf, equivalently regular holonomic D-module, with nilpotent singular support, see [Gin01]. On the spectral side there are three versions of the moduli space of local systems, all having the same complex valued points. For a unipotent group

(1.3) LocNˇσ,dR=LocNˇσ,restr=LocNˇσ,Betti\operatorname{Loc}^{\sigma,\operatorname{dR}}_{\check{N}}=\operatorname{Loc}^{\sigma,\operatorname{restr}}_{\check{N}}=\operatorname{Loc}^{\sigma,\operatorname{Betti}}_{\check{N}}

coincide by proposition 4.3.3 and section 4.8.1 of [AGK+20].

Remark.

If we replace naive Eisenstein series by compactified Eisenstein series of [BG02], then geometric Langlands predicts that Hom(Whit,Eis!K)\operatorname{Hom}(\operatorname{Whit},\operatorname{Eis}_{!*}K^{\prime}) should equal global sections of a skyscraper sheaf at σLocGˇ\sigma\in\operatorname{Loc}_{\check{G}}. This is verified by Gaitsgory in appendix B of [BHKT19].

Consistency check.

If σ\sigma is a regular then theorem 10.2 of [BG08] says that Eis!(Kλ)[dGdB0]\operatorname{Eis}_{!}(K^{\lambda})[d_{G}-d_{B}^{0}] is perverse. The Whittaker functional Hom(Whit,)[dB0]\operatorname{Hom}(\operatorname{Whit},-)[d_{B}^{0}] is exact by [NT22] or [FR22], so the automorphic side of the main theorem is concentrated in degree 0. This is consistent with LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} being a classical scheme if σ\sigma is regular. Here

(1.4) dBλ=(g1)dimB+2ρˇ,λ+2(g1)ρ=dimBunBλd_{B}^{\lambda}=(g-1)\dim B+\langle 2\check{\rho},\lambda+2(g-1)\rho\rangle=\dim\operatorname{Bun}_{B}^{\lambda}

is the dimension of the degree λ2(g1)ρ-\lambda-2(g-1)\rho connected component.

1.3. Proof outline

It is convenient to take the coweight graded linear dual to avoid topological rings and because Lie algebra homology behaves better than Lie algebra cohomology. Here is the proof of our main theorem in one sentence:

(1.5) Hom\displaystyle\operatorname{Hom} (Whit,Eis!K)[dG]=(2.1)λHom(χZDexp,qZ!DKλ)[dT+d0]\displaystyle(\operatorname{Whit},\operatorname{Eis}_{!}K)^{*}[-d_{G}]\stackrel{{\scriptstyle\eqref{ZastavaBaseChange}}}{{=}}\bigoplus_{\lambda}\operatorname{Hom}(\chi_{Z}^{*}D\exp,q_{Z}^{!}DK^{\lambda})[d_{T}+d^{0}]
=(2.4)λΓ(Xλ,Υσλ)=(2.7)Γ(Ran,C(𝔫ˇσ))=(2.10)C(Γ(X,𝔫ˇσ))=(2.11)𝒪(LocNˇσ).\displaystyle\stackrel{{\scriptstyle\eqref{PushConfiguration}}}{{=}}\bigoplus_{\lambda}\Gamma(X^{\lambda},\Upsilon^{\lambda}_{\sigma})\stackrel{{\scriptstyle\eqref{UpsilonFact}}}{{=}}\Gamma(\operatorname{Ran},C_{\bullet}(\check{\mathfrak{n}}_{\sigma}))\stackrel{{\scriptstyle\eqref{BDFormula}}}{{=}}C_{\bullet}(\Gamma(X,\check{\mathfrak{n}}_{\sigma}))\stackrel{{\scriptstyle\eqref{LieDualFunctions}}}{{=}}\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma})^{*}.

In section 2.1, we use [NT22] or [FR22] to exchange Eis!\operatorname{Eis}_{!} for a right adjoint, then apply base change and a result of [AG15] to get a calculation on the Zastava space. In section 2.2, we pushforward to the space of positive coweight valued divisors and, by theorem 4.6.1 of [Ras21], obtain a certain factorizable perverse sheaves Υσλ\Upsilon_{\sigma}^{\lambda} on XλX^{\lambda}.

In section 2.3, we interpret Υσ\Upsilon_{\sigma} in terms of the chiral enveloping algebra of 𝔫ˇσ\check{\mathfrak{n}}_{\sigma} as in [BG08]. In section 2.4, we explain, following [BG08], how the cohomology of Υσ\Upsilon_{\sigma} equals factorization homology of A=C(𝔫ˇσ)A=C_{\bullet}(\check{\mathfrak{n}}_{\sigma}). Beilinson and Drinfeld’s formula says factorization homology of C(𝔫ˇσ)C_{\bullet}(\check{\mathfrak{n}}_{\sigma}) is Lie algebra homology of Γ(X,𝔫ˇσ)\Gamma(X,\check{\mathfrak{n}}_{\sigma}). In section 2.5, we study moduli of 𝔫ˇ\check{\mathfrak{n}}-local systems using deformation theory. Since Γ(X,𝔫ˇσ)\Gamma(X,\check{\mathfrak{n}}_{\sigma}) is the shifted tangent complex of LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma}, its Lie algebra homology is related the formal completion of LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} at σ\sigma. Using that LocNˇσ=(SpecR)/Nˇ\operatorname{Loc}_{\check{N}}^{\sigma}=(\operatorname{Spec}R)/\check{N} is the quotient of an affine scheme by a unipotent group and using the contracting 𝐆m\mathbf{G}_{m}-action, we show that C(Γ(X,𝔫ˇσ))=𝒪(LocNˇσ)C_{\bullet}(\Gamma(X,\check{\mathfrak{n}}_{\sigma}))=\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma})^{*} is the graded linear dual ring of functions.

The idea of using factorization homology to study the formal completion of LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} is from [BG08]. For σ\sigma regular, propositions 11.3 and 11.4 of [BG08] give an isomorphism between Γ(Xλ,Υσλ)\prod\Gamma(X^{\lambda},\Upsilon_{\sigma}^{\lambda})^{*} and the completed ring of functions 𝒪(LocNˇσ)\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma})^{\wedge}. Sections 2.3 and 2.4 review some of their arguments and do not contain new content apart from filling in some details. Our main contribution is in section 2.5 where we extend the results of [BG08] to the more interesting case of irregular σ\sigma, and obtain a formula for the ring of functions on LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} (not just its formal completion) using the contracting 𝐆m\mathbf{G}_{m}-action.

1.4. Acknowledgements

I thank David Nadler for suggesting Whittaker coefficients of Eisenstein series and for generous discussions. This work was partially supported by NSF grant DMS-1646385.

2. Proof of the main theorem

2.1. Base change to Zastava

In this section we interpret Whittaker coefficients of Eisenstein series as twisted cohomology of the Zastava space ZZ.

The fiber product Z=BunB×BunGBunNωρZ^{\prime}=\operatorname{Bun}_{B}\times_{\operatorname{Bun}_{G}}\operatorname{Bun}_{N^{-}}^{\omega^{-\rho}} has a stratification indexed by the Weyl group, determined by the generic relative position of two flags. Let j:ZZj:Z\hookrightarrow Z^{\prime} be the open inclusion of the locus where the two flags are generically transverse, called the Zastava space.

Z{Z^{\prime}}BunB{\operatorname{Bun}_{B}}BunNωρ{\operatorname{Bun}_{N^{-}}^{\omega^{-\rho}}}BunT{\operatorname{Bun}_{T}}BunG{\operatorname{Bun}_{G}}𝐀1{\mathbf{A}^{1}}p\scriptstyle{p}q\scriptstyle{q}χ\scriptstyle{\chi}r\scriptstyle{r}

Consider the compositions

qZ:ZBunBBunTandχZ:ZBunNωρ𝐀1q_{Z^{\prime}}:Z^{\prime}\rightarrow\operatorname{Bun}_{B}\rightarrow\operatorname{Bun}_{T}\quad\text{and}\quad\chi_{Z^{\prime}}:Z^{\prime}\rightarrow\operatorname{Bun}_{N^{-}}^{\omega^{-\rho}}\rightarrow\mathbf{A}^{1}

and let qZ=qZjq_{Z}=q_{Z^{\prime}}j and χZ=χZj\chi_{Z}=\chi_{Z^{\prime}}j be their restrictions to ZZ.

Proposition 1.

There is an isomorphism

(2.1) Hom(Whit,Eis!Kλ)[dG]=Hom(χZDexp,qZ!DKλ)[dT+d0].\operatorname{Hom}(\operatorname{Whit},\operatorname{Eis}_{!}K^{\lambda})^{*}[-d_{G}]=\operatorname{Hom}(\chi_{Z}^{*}D\exp,q_{Z}^{!}DK^{\lambda})[d_{T}+d^{0}].
Proof.

We cannot directly apply adjunction to calculate Whittaker coefficients of Eisenstein series because Eis!\operatorname{Eis}_{!} is a left not right adjoint. It is shown in [FR22] and [NT22] that the shifted Whittaker functional Hom(Whit,)[dB0]\operatorname{Hom}(\operatorname{Whit},-)[d_{B}^{0}] on nilpotent sheaves commutes with Verdier duality DD. This allows us to exchange Eis!=p!q\operatorname{Eis}_{!}=p_{!}q^{*} for Eis=pq!\operatorname{Eis}_{*}=p_{*}q^{!}. Then apply adjunction and base change to reduce to a calculation on the fiber product ZZ^{\prime}.

Hom(Whit,Eis!Kλ)[2dB0]=Hom(Whit,EisDKλ)=Hom(χZDexp,qZ!DKλ)\operatorname{Hom}(\operatorname{Whit},\operatorname{Eis}_{!}K^{\lambda})^{*}[-2d_{B}^{0}]=\operatorname{Hom}(\operatorname{Whit},\operatorname{Eis}_{*}DK^{\lambda})=\operatorname{Hom}(\chi_{Z^{\prime}}^{*}D\exp,q_{Z^{\prime}}^{!}DK^{\lambda})

Finally by equation (3.5) of [AG15], restriction to the open generically transverse locus ZZ does not change the calculation. More precisely the map

Hom(χZDexp,qZ!DKλ)Hom(χZDexp,qZ!DKλ)\operatorname{Hom}(\chi_{Z^{\prime}}^{*}D\exp,q_{Z^{\prime}}^{!}DK^{\lambda})\xrightarrow{\sim}\operatorname{Hom}(\chi_{Z}^{*}D\exp,q_{Z}^{!}DK^{\lambda})

is an isomorphism. For the shifts use (1.4) and dG+dT+d0=2dB0d_{G}+d_{T}+d^{0}=2d_{B}^{0}. ∎

2.2. Pushforward to the configuration space

In this section we recall how to factor the projection qZ:ZλBunTλq_{Z}:Z^{\lambda}\rightarrow\operatorname{Bun}_{T}^{\lambda} through the configuration space XλX^{\lambda} of positive coweight valued divisors of total degree λ\lambda. Hence a description of the λ\lambda-graded piece of proposition 1 as cohomology of a certain perverse sheaf Υσλ\Upsilon^{\lambda}_{\sigma} on XλX^{\lambda}.

Let (F,F,E)Zλ(F,F^{-},E)\in Z^{\lambda} be a point in the λ\lambda connected component of Zastava space, that is a GG-bundle EE with generically transverse B,BB,B^{-}-reductions F,FF,F^{-}, such that FF has degree λ2(g1)ρ-\lambda-2(g-1)\rho and F×BT=ωρF^{-}\times_{B^{-}}T=\omega^{-\rho}. For each dominant weight μˇ\check{\mu} the Plucker description gives maps

(2.2) FμˇEμˇ(F)μˇ=ωμˇ,ρ.F^{\check{\mu}}\rightarrow E^{\check{\mu}}\rightarrow(F^{-})^{\check{\mu}}=\omega^{-\langle\check{\mu},\rho\rangle}.

Here Fμˇ=F×B𝐂μˇF^{\check{\mu}}=F\times_{B}\mathbf{C}_{\check{\mu}} is a line bundle and Eμˇ=E×GVμˇE^{\check{\mu}}=E\times_{G}V_{\check{\mu}} is the vector bundle associated to the simple GG-module of highest weight μˇ\check{\mu}.

By the generic transversality condition, the composition (2.2) is nonzero map of line bundles, so λ\lambda is a non-negative coweight. For each point in the Zastava space, there is a unique positive coweight valued divisor x¯λ¯Xλ\underline{x}\cdot\underline{\lambda}\in X^{\lambda} such that (2.2) factors through an isomorphism Fμˇ(x¯λ¯,μˇ)=ωμˇ,ρF^{\check{\mu}}(\langle\underline{x}\cdot\underline{\lambda},\check{\mu}\rangle)=\omega^{-\langle\check{\mu},\rho\rangle}. Since GG is assumed simply connected, we can write λ=niαi\lambda=\sum n_{i}\alpha_{i} as a sum of simple coroots and Xλ=X(ni)X^{\lambda}=\prod X^{(n_{i})} is a product of symmetric powers of the curve. Therefore qZq_{Z} factors through a map π\pi to the configuration space followed by the Abel-Jacobi map,

qZ:Zλ𝜋XλAJBunTλ,(E,F,F)x¯λ¯F=ωρ(x¯λ¯).q_{Z}:Z^{\lambda}\xrightarrow{\pi}X^{\lambda}\xrightarrow{\operatorname{AJ}}\operatorname{Bun}_{T}^{\lambda},\qquad(E,F,F^{-})\mapsto\underline{x}\cdot\underline{\lambda}\mapsto F=\omega^{-\rho}(-\underline{x}\cdot\underline{\lambda}).

Let λ\lambda be a coweight and n=ρˇ,λn=\langle\check{\rho},\lambda\rangle. Let 𝔫ˇσ=σ×Tˇ𝔫ˇ\check{\mathfrak{n}}_{\sigma}=\sigma\times_{\check{T}}\check{\mathfrak{n}}, an 𝔫ˇ\check{\mathfrak{n}}-local system on XX. The Chevalley complex on the coweight graded Ran space gives a Sni\prod S_{n_{i}} equivariant sheaf on Xni\prod X^{n_{i}}. Symmetrizing by pushing forward along Xni/SniXλ\prod X^{n_{i}}/S_{n_{i}}\rightarrow X^{\lambda} gives a perverse sheaf Υσλ\Upsilon_{\sigma}^{\lambda}, see section 4 of [Ras21] and equation (2.6). The definition of Υσλ\Upsilon_{\sigma}^{\lambda} involves the Chevalley differential but the associated graded of Υσλ\Upsilon_{\sigma}^{\lambda} with respect to the Cousin filtration is easier to describe, see section 3.3 of [BG08]. The stalk of Υσλ\Upsilon_{\sigma}^{\lambda} at a positive coweight valued divisor x¯λ¯Xλ\underline{x}\cdot\underline{\lambda}\in X^{\lambda} is

(2.3) (Υσλ)x¯λ¯=C(𝔫ˇσ)xiλi.(\Upsilon^{\lambda}_{\sigma})_{\underline{x}\cdot\underline{\lambda}}=\bigotimes C_{\bullet}(\check{\mathfrak{n}}_{\sigma})^{\lambda_{i}}_{x_{i}}.
Proposition 2.

There is an isomorphism

(2.4) Hom(χZDexp,qZ!DKλ)[dT+d0]=Γ(Xλ,Υσλ).\operatorname{Hom}(\chi_{Z}^{*}D\exp,q_{Z}^{!}DK^{\lambda})[d_{T}+d^{0}]=\Gamma(X^{\lambda},\Upsilon_{\sigma}^{\lambda}).
Proof.

Pushing forward to the configuration space XλX^{\lambda}, the left of (2.4) becomes

Hom(π!χZDexp,AJ!DKλ)[dT+d0]=Γ(Υλ(AJKλ))[dT+dλ]=Γ(Xλ,Υσλ).\operatorname{Hom}(\pi_{!}\chi_{Z}^{*}D\exp,\operatorname{AJ}^{!}DK^{\lambda})[d_{T}+d^{0}]\\ =\Gamma(\Upsilon^{\lambda}\otimes(\operatorname{AJ}^{*}K^{\lambda})^{*})[d_{T}+d^{\lambda}]=\Gamma(X^{\lambda},\Upsilon^{\lambda}_{\sigma}).

We used that the configuration space XλX^{\lambda} is smooth so the dualizing sheaf is a rank 1 local system. And we used theorem 4.6.1 of [Ras21], which says that

Dπ!χZDexp=πχZ!exp=Υλ[dλd0].D\pi_{!}\chi_{Z}^{*}D\exp=\pi_{*}\chi_{Z}^{!}\exp=\Upsilon^{\lambda}[d^{\lambda}-d^{0}].

Here dλd0=dimZλd^{\lambda}-d^{0}=\dim Z^{\lambda} and Υλ\Upsilon^{\lambda} is a perverse sheaf on XλX^{\lambda} with stalks Υx¯λ¯λ=C(𝔫ˇ)λi.\Upsilon^{\lambda}_{\underline{x}\cdot\underline{\lambda}}=\bigotimes C_{\bullet}(\check{\mathfrak{n}})^{\lambda_{i}}.

Under class field theory (1.1), the stalks of AJKλ\operatorname{AJ}^{*}K^{\lambda} are

(AJKλ)λ¯x¯=(σxiλi)[dT+dλ](\operatorname{AJ}^{*}K^{\lambda})_{\underline{\lambda}\cdot\underline{x}}=\left(\bigotimes\sigma^{\lambda_{i}}_{x_{i}}\right)[d_{T}+d^{\lambda}]

and its *-pullback to Xni\prod X^{n_{i}} is the Sni\prod S_{n_{i}} equivariant rank 1 local system (σαi)ni\boxtimes(\sigma^{\alpha_{i}})^{\boxtimes n_{i}}. By the projection formula, tensoring with (AJKλ)(\operatorname{AJ}^{*}K^{\lambda})^{*} has the effect of twisting Υλ\Upsilon^{\lambda} by σ\sigma. ∎

Combining propositions 1 and 2 shows Whittaker coefficients of Eisenstein series is graded dual to global sections of Υσ\Upsilon_{\sigma} on the configuration space.

2.3. The chiral enveloping algebra as a Chevalley complex

The local system 𝔫ˇσ\check{\mathfrak{n}}_{\sigma} determines a Lie* algebra on the Ran space. Its Lie algebra homology A:=C(𝔫ˇσ)A:=C_{\bullet}(\check{\mathfrak{n}}_{\sigma}) is a factorization algebra, related to Υσ\Upsilon_{\sigma} by partial symmetrization in (2.6).

A sheaf on the Ran space of XX is a collection of sheaves AXIA_{X^{I}} on each power of the curve XIX^{I} compatible under !!-restriction along all partial diagonal maps, see 4.2.1 of [BD04] or 2.1 of [FG12]. Recall from section 1.2.1 of [FG12] that the category of sheaves on the Ran space admits two tensor products with a map ch\otimes^{*}\rightarrow\otimes^{\operatorname{ch}} between them.

Pushing forward along the main diagonal Δ:XRan\Delta:X\rightarrow\operatorname{Ran}, we can regard Δ𝔫ˇσShv(Ran)\Delta_{*}\check{\mathfrak{n}}_{\sigma}\in\operatorname{Shv}(\operatorname{Ran}) as a Lie algebra for the *-tensor product. Restricting to X2X^{2}, the Lie* bracket (Δ𝔫ˇσΔ𝔫ˇσ)X2=𝔫ˇσ𝔫ˇσ(Δ𝔫ˇσ)X2=Δ𝔫ˇσ(\Delta_{*}\check{\mathfrak{n}}_{\sigma}\otimes^{*}\Delta_{*}\check{\mathfrak{n}}_{\sigma})_{X^{2}}=\check{\mathfrak{n}}_{\sigma}\boxtimes\check{\mathfrak{n}}_{\sigma}\rightarrow(\Delta_{*}\check{\mathfrak{n}}_{\sigma})_{X^{2}}=\Delta_{*}\check{\mathfrak{n}}_{\sigma} comes by adjunction from the Lie bracket.

Let A:=C(𝔫ˇσ)Shv(Ran)A:=C_{\bullet}(\check{\mathfrak{n}}_{\sigma})\in\operatorname{Shv}(\operatorname{Ran}) be Lie algebra homology of Δ𝔫ˇσ\Delta_{*}\check{\mathfrak{n}}_{\sigma} with respect to the *-tensor product, viewed by the forgetful functor as a cocommutative coalgebra with respect to the ch\operatorname{ch}-tensor product. Proposition 6.1.2 of [FG12] says that AA corresponds to the chiral enveloping algebra of Δ𝔫ˇσ\Delta_{*}\check{\mathfrak{n}}_{\sigma} under the equivalence between factorization and chiral algebras.

The Chevalley complex A=AλA=\bigoplus A^{\lambda} is coweight graded because Sym(𝔫ˇσ[1])\operatorname{Sym}(\check{\mathfrak{n}}_{\sigma}[1]) is coweight graded, and because the Chevalley differential preserves the grading. Choose a coweight λ\lambda and let n=ρˇ,λn=\langle\check{\rho},\lambda\rangle. The sheaf AXnλA^{\lambda}_{X^{n}} on XnX^{n} is SnS_{n}-equivariant and perverse. Symmetrize it along sym:XnX(n)\operatorname{sym}:X^{n}\rightarrow X^{(n)} to get a perverse sheaf (symAXnλ)Sn(\operatorname{sym}_{*}A^{\lambda}_{X^{n}})^{S_{n}} on the nnth symmetric power. (In other words we pushed forward AXnλA^{\lambda}_{X^{n}} from the stack quotient Xn/SnX^{n}/S_{n} to the coarse quotient X(n)X^{(n)}.)

Now we describe a certain perverse subsheaf AX(n)λ(symAXnλ)SnA^{\lambda}_{X^{(n)}}\subset(\operatorname{sym}_{*}A^{\lambda}_{X^{n}})^{S_{n}} defined in section 3 of [BG08]. Let Xi(n)X(n)X_{i}^{(n)}\subset X^{(n)} be the space of effective degree nn divisors supported at exactly ii points. The !!-restriction of (symAXnλ)Sn(\operatorname{sym}_{*}A^{\lambda}_{X^{n}})^{S_{n}} to Xi(n)X_{i}^{(n)} is a local system whose stalk at a divisor n¯x¯Xi(n)\underline{n}\cdot\underline{x}\in X_{i}^{(n)} is given by

(symAXinλ)n¯x¯Sn=λ=λjC(𝔫σ)xjλj.(\operatorname{sym}_{*}A^{\lambda}_{X^{n}_{i}})^{S_{n}}_{\underline{n}\cdot\underline{x}}=\bigoplus_{\lambda=\sum\lambda_{j}}\bigotimes C_{\bullet}(\mathfrak{n}_{\sigma})^{\lambda_{j}}_{x_{j}}.

The !!-restriction of AX(n)λA^{\lambda}_{X^{(n)}} to Xi(n)X(n)X^{(n)}_{i}\subset X^{(n)} is the summand whose stalks are

(2.5) (AXi(n)λ)n¯x¯=λ=λj,ρˇ,λj=njC(𝔫ˇσ)xjλj.(A^{\lambda}_{X^{(n)}_{i}})_{\underline{n}\cdot\underline{x}}=\bigoplus_{\begin{subarray}{c}\lambda=\sum\lambda_{j},\\ \langle\check{\rho},\lambda_{j}\rangle=n_{j}\end{subarray}}\bigotimes C_{\bullet}(\check{\mathfrak{n}}_{\sigma})_{x_{j}}^{\lambda_{j}}.

By section 11.6 of [BG08], the pushforward of Υσλ\Upsilon_{\sigma}^{\lambda}, see (2.3), along the partial symmetrization map symλ:XλX(n)\operatorname{sym}^{\lambda}:X^{\lambda}\rightarrow X^{(n)} is

(2.6) symλΥσλ=AX(n)λ.\operatorname{sym}^{\lambda}_{*}\Upsilon_{\sigma}^{\lambda}=A^{\lambda}_{X^{(n)}}.

2.4. Factorization homology

In this section we review, following [BG08], how factorization homology of Aλ=C(𝔫ˇσ)λA^{\lambda}=C_{\bullet}(\check{\mathfrak{n}}_{\sigma})^{\lambda} can be computed as cohomology on the symmetric power X(n)X^{(n)}, where n=ρˇ,λn=\langle\check{\rho},\lambda\rangle.

Let FSet\operatorname{FSet} be the category whose objects are finite nonempty sets and whose morphisms are surjective maps. For each surjection JIJ\twoheadrightarrow I there is a partial diagonal map Δ:XIXJ\Delta:X^{I}\rightarrow X^{J}. By definition of a sheaf on the Ran space AXI=Δ!AXJA_{X^{I}}=\Delta^{!}A_{X^{J}} so adjunction gives maps ΔAXIAXJ\Delta_{*}A_{X^{I}}\rightarrow A_{X^{J}}. Factorization homology is defined in section 6.3.3 of [FG12] or section 4.2.2 of [BD04] as the colimit over these maps

Γ(Ran,A)=colimFSetopΓ(AXI).\Gamma(\operatorname{Ran},A)=\underset{\operatorname{FSet}^{\operatorname{op}}}{\operatorname{colim}}\Gamma(A_{X^{I}}).

The following proposition is stated in 11.6 of [BG08] and below we fill in the proof using the Cousin filtration and ideas from section 4.2 of [BD04].

Proposition 3.

The cohomology of Υσ\Upsilon_{\sigma}, see (2.3), is the factorization homology of the Chevalley complex,

(2.7) λΓ(Xλ,Υσλ)=Γ(Ran,A).\bigoplus_{\lambda}\Gamma(X^{\lambda},\Upsilon_{\sigma}^{\lambda})=\Gamma(\operatorname{Ran},A).
Proof.

Equation (2.6) relates Υσ\Upsilon_{\sigma} to the symmetrization of AA. Thus it suffices to show that

(2.8) Γ(Xλ,Υσλ)=Γ(AX(n)λ)Γ(AXnλ)Γ(Ran,Aλ)\Gamma(X^{\lambda},\Upsilon_{\sigma}^{\lambda})=\Gamma(A^{\lambda}_{X^{(n)}})\rightarrow\Gamma(A^{\lambda}_{X^{n}})\rightarrow\Gamma(\operatorname{Ran},A^{\lambda})

is an isomorphism for n=ρˇ,λn=\langle\check{\rho},\lambda\rangle. Indeed we will prove that (2.8) is compatible with the Cousin filtration and that it induces an isomorphism on the associated graded pieces.

Consider the filtration on (2.8) whose i\leq ith filtered piece consists of sections supported on the partial diagonals of dimensions i\leq i. The iith graded piece is

(2.9) Γ(AXi(n)λ)Γ(AXinλ)colimFSetopΓ(AXiIλ)=griΓ(Ran,Aλ).\Gamma(A^{\lambda}_{X^{(n)}_{i}})\rightarrow\Gamma(A^{\lambda}_{X^{n}_{i}})\rightarrow\underset{\operatorname{FSet}^{\operatorname{op}}}{\operatorname{colim}}\Gamma(A^{\lambda}_{X^{I}_{i}})=\operatorname{gr}_{i}\Gamma(\operatorname{Ran},A^{\lambda}).

Here AXi(n)λA^{\lambda}_{X_{i}^{(n)}} is the !-restriction of AX(n)λA^{\lambda}_{X^{(n)}} to the space Xi(n)X(n)X_{i}^{(n)}\subset X^{(n)} of effective degree nn divisors supported at exactly ii points. Similarly AXiIλA^{\lambda}_{X_{i}^{I}} is the !-restriction of AXIλA^{\lambda}_{X^{I}} to the space XiIXIX_{i}^{I}\subset X^{I} of II-tuples supported at exactly ii points.

The symmetric group SiS_{i} acts freely on the space XiiXiX^{i}_{i}\subset X^{i} of distinct ii-tuples of points. By section 4.2.3 of [BD04], the iith graded piece of the factorization homology of AλA^{\lambda} is griΓ(Ran,Aλ)=Γ(AXiiλ)Si\operatorname{gr}_{i}\Gamma(\operatorname{Ran},A^{\lambda})=\Gamma(A^{\lambda}_{X^{i}_{i}})_{S_{i}}.

The connected components of Xi(n)X^{(n)}_{i} are indexed by partitions n¯=n1+ni\underline{n}=n_{1}+\dots n_{i}. Also the local system AXiiλA^{\lambda}_{X^{i}_{i}} splits as a direct sum indexed by such partitions. Restricting (2.9) to the connected component Xi(n¯)Xi(n)X^{(\underline{n})}_{i}\subset X^{(n)}_{i} indexed by a certain partition,

Γ(AXi(n¯)λ)griΓ(Ran,Aλ)=Γ(AXiiλ)Si\Gamma(A^{\lambda}_{X^{(\underline{n})}_{i}})\rightarrow\operatorname{gr}_{i}\Gamma(\operatorname{Ran},A^{\lambda})=\Gamma(A^{\lambda}_{X^{i}_{i}})_{S_{i}}

is an isomorphism onto the corresponding summand of Γ(AXiiλ)Si\Gamma(A^{\lambda}_{X^{i}_{i}})_{S_{i}} by (2.5). Summing over partitions shows that the iith graded piece of (2.8) is an isomorphism. ∎

Since the factorization algebra A=C(𝔫ˇσ)A=C_{\bullet}(\check{\mathfrak{n}}_{\sigma}) corresponds to the chiral enveloping algebra U(𝔫ˇσ)U(\check{\mathfrak{n}}_{\sigma}), Beilinson and Drinfeld’s formula for chiral homology of an enveloping algebra, see theorem 4.8.1.1 of [BD04] or 6.4.4 of [FG12], says

(2.10) Γ(Ran,A)=C(Γ(X,𝔫ˇσ)).\Gamma(\operatorname{Ran},A)=C_{\bullet}(\Gamma(X,\check{\mathfrak{n}}_{\sigma})).

2.5. Deformation theory

In this section we show that

(2.11) C(Γ(X,𝔫ˇσ))=𝒪(LocNˇσ),C_{\bullet}(\Gamma(X,\check{\mathfrak{n}}_{\sigma}))=\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma})^{*},

Lie algebra homology of the shifted tangent complex equals the graded dual ring of functions on LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma}. Deformation theory says that C(Γ(X,𝔫ˇσ))=ΓIndCoh(ω(LocNˇσ))C_{\bullet}(\Gamma(X,\check{\mathfrak{n}}_{\sigma}))=\Gamma^{\operatorname{IndCoh}}(\omega_{(\operatorname{Loc}_{\check{N}}^{\sigma})^{\wedge}}) is global sections of the dualizing sheaf on the formal completion at σ\sigma. Using the structure of LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma} described in proposition 4, we can recover the graded dual ring of functions on LocNˇσ\operatorname{Loc}_{\check{N}}^{\sigma}, not just its completion, from ΓIndCoh(ω(LocNˇσ))\Gamma^{\operatorname{IndCoh}}(\omega_{(\operatorname{Loc}_{\check{N}}^{\sigma})^{\wedge}}).

First we show that LocNˇσ=LocNˇσ,x/Nˇ\operatorname{Loc}_{\check{N}}^{\sigma}=\operatorname{Loc}_{\check{N}}^{\sigma,x}/\check{N} is the quotient by a unipotent group of an affine derived scheme with a contracting 𝐆m\mathbf{G}_{m}-action. Let LocBˇx=Bˇ2g×Bˇ1\operatorname{Loc}_{\check{B}}^{x}=\check{B}^{2g}\times_{\check{B}}1 (respectively LocTˇx=Tˇ2g×Tˇ1\operatorname{Loc}_{\check{T}}^{x}=\check{T}^{2g}\times_{\check{T}}1) be the Betti moduli of Bˇ\check{B} (respectively Tˇ\check{T}) local systems trivialized at a point xx. Let LocNˇσ,x=LocBˇx×LocTˇxσ\operatorname{Loc}_{\check{N}}^{\sigma,x}=\operatorname{Loc}_{\check{B}}^{x}\times_{\operatorname{Loc}_{\check{T}}^{x}}\sigma be the moduli of Bˇ\check{B}-local systems with underlying Tˇ\check{T}-local system identified with σ\sigma, plus a Tˇ\check{T}-reduction at xx.

Since Tˇ\check{T} is abelian, it acts by automorphisms on σLocTˇ\sigma\in\operatorname{Loc}_{\check{T}} so there is a canonical lift σLocTˇx\sigma\in\operatorname{Loc}_{\check{T}}^{x}. We also sometimes regard σ\sigma as a point in LocNˇσ,x\operatorname{Loc}_{\check{N}}^{\sigma,x} via the inclusion TˇBˇ\check{T}\subset\check{B}.

Let Bˇ\check{B} act on LocBˇx\operatorname{Loc}_{\check{B}}^{x} by changing the trivialization at xx, equivalently by the adjoint action on Bˇ2g×Bˇ1\check{B}^{2g}\times_{\check{B}}1. Restricting the adjoint action along ρˇ\check{\rho} gives a 𝐆m\mathbf{G}_{m}-action that contracts Bˇ\check{B} to Tˇ\check{T}. Thus we expect a 𝐆m\mathbf{G}_{m}-action that contracts LocBˇx\operatorname{Loc}_{\check{B}}^{x} to LocTˇx\operatorname{Loc}_{\check{T}}^{x}, as is made precise below.

Proposition 4.

The moduli space LocNˇσ,x=SpecR\operatorname{Loc}_{\check{N}}^{\sigma,x}=\operatorname{Spec}R is a finite type affine scheme with a Bˇ\check{B}-action. Restricting the action along ρˇ\check{\rho} gives a non-negative grading R=n0RnR=\bigoplus_{n\geq 0}R_{n} such that σ=SpecR/R>0\sigma=\operatorname{Spec}R/R_{>0} is cut out by the ideal of strictly positively graded functions.

Proof.

We argue in the Betti setting, but the restricted and de Rham versions also follow by (1.3). First rewrite

(2.12) LocNˇσ,x=LocBˇx×LocTˇxσ=Bˇ2g×Tˇ2g×TˇBˇσ=(Bˇ2g×Tˇ2gσ)×Bˇ×Tˇ11=Spec(RSk).\operatorname{Loc}_{\check{N}}^{\sigma,x}=\operatorname{Loc}_{\check{B}}^{x}\times_{\operatorname{Loc}_{\check{T}}^{x}}\sigma=\check{B}^{2g}\times_{\check{T}^{2g}\times_{\check{T}}\check{B}}\sigma=(\check{B}^{2g}\times_{\check{T}^{2g}}\sigma)\times_{\check{B}\times_{\check{T}}1}1=\operatorname{Spec}(R^{\prime}\otimes_{S}k).

The contracting ρˇ\check{\rho}-action induces non-negative gradings on the classical rings R=𝒪(Bˇ2g×Tˇ2gσ)R^{\prime}=\mathcal{O}(\check{B}^{2g}\times_{\check{T}^{2g}}\sigma) and S=𝒪(Bˇ×Tˇ1)=𝒪(Nˇ)S=\mathcal{O}(\check{B}\times_{\check{T}}1)=\mathcal{O}(\check{N}). Since Nˇ\check{N} is smooth, the augmentation module k=S/S>0k=S/S_{>0} admits a finite graded resolution by free SS-modules, with all but one term shifted into strictly positive ρˇ\check{\rho}-gradings. Therefore R=RSkR=R^{\prime}\otimes_{S}k is a finite type non-negatively graded ring and σ=SpecR/R>0\sigma=\operatorname{Spec}R/R_{>0}. ∎

Now we review some derived deformation theory. Let YY^{\wedge} be the formal completion of a derived stack YY at a point σ\sigma. The shifted tangent bundle TσY[1]T_{\sigma}Y[-1] is a dg Lie algebra whose enveloping algebra is endomorphisms of the skyscraper at σ\sigma. By chapter 7 of [GR17b] or remark 2.4.2 of [Lur11], there is an equivalence

Mod(TσY[1])=IndCoh(Y)\operatorname{Mod}(T_{\sigma}Y[-1])=\operatorname{IndCoh}(Y^{\wedge})

between Lie algebra modules for the shifted tangent complex and indcoherent sheaves on the formal completion. Let p:Yptp:Y^{\wedge}\rightarrow\operatorname{pt} be the map to a point. By chapter 7 section 5.2 of [GR17b], the trivial TσY[1]T_{\sigma}Y[-1]-module corresponds to the dualizing sheaf ωY=p!kIndCoh(Y)\omega_{Y^{\wedge}}=p^{!}k\in\operatorname{IndCoh}(Y^{\wedge}). Moreover Lie algebra homology corresponds to global sections

(2.13) C(TσY[1])=ΓIndCoh(ωY).C_{\bullet}(T_{\sigma}Y[-1])=\Gamma^{\operatorname{IndCoh}}(\omega_{Y^{\wedge}}).

Suppose Y=SpecRY^{\wedge}=\operatorname{Spec}R is the spectrum of an Artinian local ring RR. By properness, p!p^{!} is right adjoint to pIndCohp_{*}^{\operatorname{IndCoh}}. Therefore the dualizing complex ωY=R\omega_{Y^{\wedge}}=R^{*} is the linear dual of RR viewed as an RR-module.

Suppose Y=SpfR=colimYnY^{\wedge}=\operatorname{Spf}R^{\wedge}=\operatorname{colim}Y_{n} where Yn=SpecR/𝔪nY_{n}=\operatorname{Spec}R/\mathfrak{m}^{n} and let in:YnYi_{n}:Y_{n}\rightarrow Y^{\wedge}. Since YnYn+1Y_{n}\rightarrow Y_{n+1} is proper, IndCoh(Y)\operatorname{IndCoh}(Y^{\wedge}) is the colimit under *-pushforward of IndCoh(Yn)\operatorname{IndCoh}(Y_{n}), see chapter 1 proposition 2.5.7 of [GR17a]. The dualizing sheaf can be written as a colimit, ωY=coliminIndCohωYn\omega_{Y^{\wedge}}=\operatorname{colim}i_{n*}^{\operatorname{IndCoh}}\omega_{Y_{n}}, see chapter 7 corollary 5.3.3 of [GR17b]. Since ΓIndCoh(Y,)\Gamma^{\operatorname{IndCoh}}(Y^{\wedge},-) is continuous it follows that

(2.14) ΓIndCoh(ωY)=colim((R/𝔪n))=(R)\Gamma^{\operatorname{IndCoh}}(\omega_{Y^{\wedge}})=\operatorname{colim}((R/\mathfrak{m}^{n})^{*})=(R^{\wedge})^{*}

is the topological dual of the completed local ring RR^{\wedge}. In this case, equation (2.13) is corollary 5.2 of [Hin96].

Proposition 5.

Let R=n0RnR=\bigoplus_{n\geq 0}R_{n} be a non-negatively graded finite type derived ring with R0=kR_{0}=k. Let RR^{\wedge} be the formal completion with respect to the ideal of positively graded functions. Then the graded dual R=RnR^{*}=\bigoplus R_{n}^{*} equals the topological dual of the completion (R)(R^{\wedge})^{*}.

Proof.

First suppose RR is classical and choose homogeneous generators f1,frRf_{1},\dots f_{r}\in R. Let dd be the maximum of their degrees, so Rdn(f1,fr)nRnR_{\geq dn}\subset(f_{1},\dots f_{r})^{n}\subset R_{\geq n}. Therefore the graded dual RR^{*} (linear functionals that vanish on some RnR_{\geq n}) equals the topologogical dual (R)(R^{\wedge})^{*} (linear functionals that vanish on some (f1,fr)n(f_{1},\dots f_{r})^{n}).

Now suppose that RR is derived. The finite type assumption means that after taking cohomology H(R)H^{\bullet}(R) is a finitely generated module over H0(R)H^{0}(R), a finitely generated graded classical ring. Choose a finite collection of homogeneous elements f1,frRf_{1},\dots f_{r}\in R whose images generate H0(R)H^{0}(R).

The formal completion is the topological ring

R=Rk[f1,fr]k[[f1,fr]]=limnRk[f1,fr](k[f1,fr]/k[f1,fr]>n).R^{\wedge}=R\otimes_{k[f_{1},\dots f_{r}]}k[[f_{1},\dots f_{r}]]=\lim_{n}R\otimes_{k[f_{1},\dots f_{r}]}(k[f_{1},\dots f_{r}]/k[f_{1},\dots f_{r}]_{>n}).

For the first equality, see section 6.7 of [GR14]. The second equality uses that fiber products commute with filtered colimits and that k[[f1,fr]]=lim(k[f1,fr]/k[f1,fr]>n)k[[f_{1},\dots f_{r}]]=\lim(k[f_{1},\dots f_{r}]/k[f_{1},\dots f_{r}]_{>n}). (The formal completion of a classical positively graded polynomial algebra can be computed using the grading filtration.)

Since k[f1,fr]k[f_{1},\dots f_{r}] is smooth, Rk[f1,fr]k[f1,fr]/k[f1,fr]>nR\otimes_{k[f_{1},\dots f_{r}]}k[f_{1},\dots f_{r}]/k[f_{1},\dots f_{r}]_{>n} has finite dimensional cohomology and therefore is concentrated in bounded degrees. Hence for mm sufficiently large the quotient map factors through

RR/R>mRk[f1,fr]k[f1,fr]/k[f1,fr]>nR/R>n.R\rightarrow R/R_{>m}\rightarrow R\otimes_{k[f_{1},\dots f_{r}]}k[f_{1},\dots f_{r}]/k[f_{1},\dots f_{r}]_{>n}\rightarrow R/R_{>n}.

Therefore the formal completion of RR can be computed using the grading filtration

R=limnRk[f1,fr](k[f1,fr]/k[f1,fr]>n)=limnR/R>n.R^{\wedge}=\lim_{n}R\otimes_{k[f_{1},\dots f_{r}]}(k[f_{1},\dots f_{r}]/k[f_{1},\dots f_{r}]_{>n})=\lim_{n}R/R_{>n}.

Taking the topological dual proves (R)=colim((R/R>n))=Rn=R(R^{\wedge})^{*}=\operatorname{colim}((R/R_{>n})^{*})=\bigoplus R_{n}^{*}=R^{*}. ∎

The following proposition shows (2.11), completing the final step of (1.5) and the proof of the main theorem.

Proposition 6.

Lie algebra homology of the shifted tangent complex of Y=LocNˇσY=\operatorname{Loc}_{\check{N}}^{\sigma} equals the graded dual of the ring of functions,

C(TσY[1])=𝒪(Y).C_{\bullet}(T_{\sigma}Y[-1])=\mathcal{O}(Y)^{*}.
Proof.

Write LocNˇσ,x=SpecR\operatorname{Loc}_{\check{N}}^{\sigma,x}=\operatorname{Spec}R as in proposition 4. Let Nˇ\check{N} act by changing the Tˇ\check{T}-reduction at xx. Since Tˇ\check{T} normalizes Nˇ\check{N}, the quotient Y=LocNˇσ=(SpecR)/NˇY=\operatorname{Loc}_{\check{N}}^{\sigma}=(\operatorname{Spec}R)/\check{N} retains the ρˇ\check{\rho}-action. The formal completion of YY at σ\sigma is the inf-scheme Y=Spf(R)/exp(𝔫ˇ)Y^{\wedge}=\operatorname{Spf}(R^{\wedge})/\exp(\check{\mathfrak{n}}), the quotient by the formal group exp(𝔫ˇ)\exp(\check{\mathfrak{n}}).

Deformation theory says

C(TσY[1])=ΓIndCoh(ωY)=((R))𝔫ˇ.C_{\bullet}(T_{\sigma}Y[-1])=\Gamma^{\operatorname{IndCoh}}(\omega_{Y^{\wedge}})=((R^{\wedge})^{*})_{\check{\mathfrak{n}}}.

The first equality is equation (2.13). For the second equality we pushed forward the dualizing sheaf ωY\omega_{Y^{\wedge}} in two steps,

Ypt/exp(𝔫ˇ)pt.Y^{\wedge}\rightarrow\operatorname{pt}/\exp(\check{\mathfrak{n}})\rightarrow\operatorname{pt}.

The pushforward of ωY\omega_{Y^{\wedge}} to pt/exp(𝔫ˇ)\operatorname{pt}/\exp(\check{\mathfrak{n}}) is an 𝔫ˇ\check{\mathfrak{n}}-module. By proper base change and (2.14), the underlying vector space is ΓIndCoh(ωSpfR)=(R)\Gamma^{\operatorname{IndCoh}}(\omega_{\operatorname{Spf}R^{\wedge}})=(R^{\wedge})^{*} and the 𝔫ˇ\check{\mathfrak{n}}-module structure comes from the Nˇ\check{N}-action. Further pushing forward along pt/exp(𝔫ˇ)pt\operatorname{pt}/\exp(\check{\mathfrak{n}})\rightarrow\operatorname{pt} corresponds to taking 𝔫ˇ\check{\mathfrak{n}}-coinvariants so ΓIndCoh(ωY)=((R))𝔫ˇ\Gamma^{\operatorname{IndCoh}}(\omega_{Y^{\wedge}})=((R^{\wedge})^{*})_{\check{\mathfrak{n}}}.

Now we show that 𝔫ˇ\check{\mathfrak{n}}-coinvariants of the topological dual of RR^{\wedge} equals the graded dual ring of functions on YY,

((R))𝔫ˇ=colim(((R/R>n))𝔫ˇ)=colim(((R/R>n)𝔫ˇ))=colim(((R𝔫ˇ/(R𝔫ˇ)>n))=(RNˇ).((R^{\wedge})^{*})_{\check{\mathfrak{n}}}=\operatorname{colim}(((R/R_{>n})^{*})_{\check{\mathfrak{n}}})=\operatorname{colim}(((R/R_{>n})^{\check{\mathfrak{n}}})^{*})=\operatorname{colim}(((R^{\check{\mathfrak{n}}}/(R^{\check{\mathfrak{n}}})_{>n})^{*})=(R^{\check{N}})^{*}.

The ideal R>nR_{>n} is an 𝔫ˇ\check{\mathfrak{n}}-module because the 𝔫ˇ\check{\mathfrak{n}}-action increases ρˇ\check{\rho}-weights. For the first equality, proposition 5 says that (R)=colim((R/R>n))(R^{\wedge})^{*}=\operatorname{colim}((R/R_{>n})^{*}), and coinvariants commutes with colimits. For the second equality, ((R/R>n))𝔫ˇ=((R/R>n)𝔫ˇ)((R/R_{>n})^{*})_{\check{\mathfrak{n}}}=((R/R_{>n})^{\check{\mathfrak{n}}})^{*} because R/R>nR/R_{>n} has finite dimensional cohomology. For the third equality, the image of (R>n)𝔫ˇR𝔫ˇ(R_{>n})^{\check{\mathfrak{n}}}\rightarrow R^{\check{\mathfrak{n}}} is concentrated in degrees >n>n so we get a map (R/R>n)𝔫ˇR𝔫ˇ/(R𝔫ˇ)>n(R/R_{>n})^{\check{\mathfrak{n}}}\rightarrow R^{\check{\mathfrak{n}}}/(R^{\check{\mathfrak{n}}})_{>n}. Moreover since (R/R>n)𝔫ˇ(R/R_{>n})^{\check{\mathfrak{n}}} is concentrated in bounded degrees, for mm sufficiently large the quotient map factors through

R𝔫ˇ/(R𝔫ˇ)>m(R/R>n)𝔫ˇR𝔫ˇ/(R𝔫ˇ)>n.R^{\check{\mathfrak{n}}}/(R^{\check{\mathfrak{n}}})_{>m}\rightarrow(R/R_{>n})^{\check{\mathfrak{n}}}\rightarrow R^{\check{\mathfrak{n}}}/(R^{\check{\mathfrak{n}}})_{>n}.

For the fourth equality, we used the van Est isomorphism, see theorem 5.1 of [Hoc61]. Since Nˇ\check{N} is unipotent, Lie algebra cohomology R𝔫ˇR^{\check{\mathfrak{n}}} coincides with group cohomology RNˇR^{\check{N}}. ∎

Example.

Let G=SL(2)G=\operatorname{SL}(2) and let σ\sigma be a Tˇ\check{T}-local system, viewed as a rank 1 local system using the positive coroot. Then σ\sigma is regular if and only if it is nontrivial.

If σ\sigma is regular, then LocNˇσ=H1(X,σ)\operatorname{Loc}_{\check{N}}^{\sigma}=H^{1}(X,\sigma) is a classical affine scheme because the other cohomologies vanish. The shifted tangent complex TσLocNˇσ[1]=H1(X,σ)[1]T_{\sigma}\operatorname{Loc}_{\check{N}}^{\sigma}[-1]=H^{1}(X,\sigma)[-1] is an abelian Lie algebra with enveloping algebra U=Sym(H1(X,σ)[1])U=\operatorname{Sym}(H^{1}(X,\sigma)[-1]). Lie algebra homology of the shifted tangent complex is

kUk=SymH1(X,σ)=𝒪(LocNˇσ).k\otimes_{U}k=\operatorname{Sym}H^{1}(X,\sigma)=\mathcal{O}(\operatorname{Loc}_{\check{N}}^{\sigma})^{*}.

If σ\sigma is trivial then C(TσLocNˇ[1])=Sym(H2(X)[1]H1(X)H0(X)[1])C_{\bullet}(T_{\sigma}\operatorname{Loc}_{\check{N}}[-1])=\operatorname{Sym}(H^{2}(X)[-1]\oplus H^{1}(X)\oplus H^{0}(X)[1]) is the graded dual ring of functions on LocNˇ=H2(X)[1]×H1(X)×pt/H0(X)\operatorname{Loc}_{\check{N}}=H^{2}(X)[-1]\times H^{1}(X)\times\operatorname{pt}/H^{0}(X).

References

  • [AG15] Dima Arinkin and Dennis Gaitsgory. Asymptotics of geometric Whittaker coefficients. Preprint. https://people.mpim-bonn.mpg.de/gaitsgde/GL/WhitAsympt.pdf, 2015.
  • [AGK+20] Dima Arinkin, Dennis Gaitsgory, David Kazhdan, Sam Raskin, Nick Rozenblyum, and Yasha Varshavsky. The stack of local systems with restricted variation and geometric Langlands theory with nilpotent singular support. arXiv preprint arXiv:2010.01906, 2020.
  • [BD04] Alexander Beilinson and Vladimir Drinfeld. Chiral algebras, volume 51 of American mathematical society colloquium publications. American Mathematical Society, Providence, RI, 1:25–26, 2004.
  • [BG02] Alexander Braverman and Dennis Gaitsgory. Geometric Eisenstein series. Invent. math, 150:287–384, 2002.
  • [BG08] Alexander Braverman and Dennis Gaitsgory. Deformations of local systems and Eisenstein series. GAFA Geom. funct. anal, 17:1788–1850, 2008.
  • [BHKT19] Gebhard Böckle, Michael Harris, Chandrashekhar Khare, and Jack A Thorne. G-local systems on smooth projective curves are potentially automorphic. Acta Mathematica, 223(1):1–111, 2019.
  • [FG12] John Francis and Dennis Gaitsgory. Chiral Koszul duality. Selecta Mathematica, 18(1):27–87, 2012.
  • [FGV01] Edward Frenkel, Dennis Gaitsgory, and Kari Vilonen. Whittaker patterns in the geometry of moduli spaces of bundles on curves. Annals of Mathematics, 153(3):699–748, 2001.
  • [FR22] Joakim Færgeman and Sam Raskin. Non-vanishing of geometric Whittaker coefficients for reductive groups. 2022.
  • [Gai10] Dennis Gaitsgory. Notes on geometric Langlands: The extended Whittaker category, 2010.
  • [Gai13] Dennis Gaitsgory. Outline of the proof of the geometric Langlands conjecture for GL(2). arXiv preprint arXiv:1302.2506, 2013.
  • [Gin01] Victor Ginzburg. The global nilpotent variety is Lagrangian. Duke Mathematical Journal, 109(3):511–519, 2001.
  • [GR14] Dennis Gaitsgory and Nick Rozenblyum. DG indschemes. Perspectives in representation theory, 610:139–251, 2014.
  • [GR17a] Dennis Gaitsgory and Nick Rozenblyum. A Study in Derived Algebraic Geometry: Volume I: Correspondences and Duality, volume 221. 2017.
  • [GR17b] Dennis Gaitsgory and Nick Rozenblyum. A Study in Derived Algebraic Geometry: Volume II: Deformations, Lie Theory and Formal Geometry, volume 221. 2017.
  • [Hin96] Vladimir Hinich. Descent of Deligne groupoids. arXiv preprint alg-geom/9606010, 1996.
  • [Hoc61] Gerhard Hochschild. Cohomology of algebraic linear groups. Illinois Journal of Mathematics, 5(3):492–519, 1961.
  • [Lur11] Jacob Lurie. DAG X: Formal moduli problems. http://www.math.harvard.edu/ lurie, 2011.
  • [NT22] David Nadler and Jeremy Taylor. The Whittaker functional is a shifted microstalk. arXiv preprint arXiv:2206.09216, 2022.
  • [NY19] David Nadler and Zhiwei Yun. Geometric Langlands correspondence for SL(2), PGL(2) over the pair of pants. Compositio Mathematica, 155(2):324–371, 2019.
  • [Ras21] Sam Raskin. Chiral principal series categories I: finite dimensional calculations. Advances in Mathematics, 388:107856, 2021.