Affine Beilinson-Bernstein localization at the critical level for GL₂

From MaRDI portal
Publication:2078804



Abstract: We prove the rank 1 case of a conjecture of Frenkel-Gaitsgory: critical level Kac-Moody representations with regular central characters localize onto the affine Grassmannian. The method uses an analogue in local geometric Langlands of the existence of Whittaker models for most representations of GL2 over a non-Archimedean field.


The paper under review proves an important special case of a conjecture of Frenkel and Gaitsgory on an affine analogue of the Beilinson-Bernstein localization theorem. To provide a context of the main theorem, let us first discuss the Beilinson--Bernstein theorem. Let \(G\) be a reductive group, \(B\) be a Borel subgroup, and \(\mathfrak{g}\) and \(\mathfrak{b}\) be their Lie algebras, respectively. One of the central problems in the subject of representation theory is to classify representations of \(\mathfrak{g}\). As the center \(Z(\mathfrak{g})\) of the universal enveloping algebra \(U(\mathfrak g)\) acts through a character, called a central character, one may as well fix the central character. Then one is led to the study of the category \(\mathfrak g\text{-mod}_0\) of representations of \(U(\mathfrak{g})\) with the central character being the same as the trivial representation. Consider the flag variety \(G/B\) and the category \(D(G/B)\) of D-modules on it. For an object \(M \in D(G/B)\), its space \(\Gamma (G/B,M)\) of global sections has an induced action of \(\mathfrak{g}\) and one can check that it has the same central character as the trivial representation. Then the celebrated Beilinson-Bernstein localization theorem says that the functor \[ \Gamma \colon D(G/B)\to \mathfrak{g}\text{-mod}_0 \] is in fact an equivalence of abelian and derived categories. For an affine analogue, consider an affine Kac-Moody algebra \(\widehat{\mathfrak{g}}_\kappa \) where \(\kappa\) is a level, or a symmetric invariant bilinear form on \(\mathfrak{g}\). Feigin and Frenkel proved that the center \(Z(\widehat{\mathfrak{g}}_{\kappa }):= Z(U(\widehat{\mathfrak{g}}_\kappa ))\) of the universal enveloping algebra \(U(\widehat{\mathfrak{g}}_\kappa )\) is trivial unless \(\kappa\) is the critical level. Moreover, they identified the center at the critical level in terms of the Langlands dual group \(\check{G}\) and exhibited the following commutative diagrams: \begin{center} \begin{tikzpicture} \draw [-, thick] (0,1.5) to (0,0.5) ; \draw [-, thick] (3,1.5) to (3,0.5) ; \draw [->>, thick] (1,2) to (2,2) ; \draw [->>, thick] (1,0) to (2,0) ; \node at (-0.5,1) {\( \cong \)}; \node at (3.5,1) {\( \cong \)}; \node at (0,0) {\(\text{Fun}(\text{Op}_{\check{G}} )\)}; \node at (3,0) {\(\text{Fun}( \text{Op}_{\check{G}}^{\text{reg}})\)}; \node at (0,2) {\(Z(\widehat{\mathfrak{g}}_{\text{crit}})\)}; \node at (3,2) {\(\text{End}(\mathbb{V}_{\text{crit}})\)}; \end{tikzpicture} \end{center} where \(\mathbb{V}_{\text{crit}} := \text{Ind}^{\widehat{\mathfrak{g}}_{\text{crit}} }_{\mathfrak gt}\mathbb C\) is the vacuum module, \( \text{Op}_{\check{G}}\) is the space of \(\check{G}\)-opers on the punctured disk, and \( \text{Op}_{\check{G}}^{\text{reg}}\) is the space of \(\check{G}\)-opers on the disk. Now we are interested in understanding the category \( \widehat{\mathfrak{g}}_{\text{crit}}\text{-mod}_{\text{reg}} \) of \(\widehat{\mathfrak{g}}_{\text{crit}} \)-modules where \(Z(\widehat{\mathfrak{g}}_{\text{crit}} )\) acts through the quotient \(\text{Fun}( \text{Op}_{\check{G}}^{\text{reg}})\) of \(\text{Fun}(\text{Op}_{\check{G}} )\). Just like the finite-dimensional case, we have a functor \[ \Gamma \colon D_{\text{crit}}(\text{Gr}_G) \to \widehat{\mathfrak{g}}_{\text{crit}}\text{-mod}_{\text{reg}} , \] where \(\text{Gr}_G:=G(K)/G(O)\) is the affine Grassmannian of \(G\) understood as a flag variety of the loop group \(G(K)\). However, this has no chance of being an equivalence; the delta function \(\delta_1\) goes to the vacuum module \(\mathbb{V}_{\text{crit}}\), but the endomorphisms of the delta function are trivial whereas the endomorphisms of the vacuum module are big, as mentioned before. Frenkel and Gaitsgory's paper ``Local geometric Langlands correspondence and affine Kac-Moody algebras suggested to instead consider the following functor \[ \Gamma^{\text{Hecke}} \colon D_{\text{crit}}(\text{Gr}_G) \otimes_{\text{Rep}(\check{G})} \text{QCoh}( \text{Op}_{\check{G}}^{\text{reg}}) \to \widehat{\mathfrak{g}}_{\text{crit}}\text{-mod}_{\text{reg}} \] where \begin{itemize} \item the geometric Satake functor \(\text{Rep}(\check{G}) \to D_{\text{crit}}(\text{Gr}_G)^{G(O)}\) is a symmetric monoidal functor and \(D_{\text{crit}}(\text{Gr}_G)^{G(O)}\) acts on \(D_{\text{crit}}(\text{Gr}_G)\); \item as the space \( \text{Op}_{\check{G}}^{\text{reg}}\) is the moduli space of flat \(\check{G}\)-bundles on the formal disk with an extra structure, the forgetful map \( \text{Op}_{\check{G}}^{\text{reg}}\to \text{Flat}_{\check{G}}(D) = B\check{G} \) gives a symmetric monoidal functor \(\text{Rep}(\check{G})\to \text{QCoh}( \text{Op}_{\check{G}}^{\text{reg}})\). \end{itemize} Then their conjecture is that \(\Gamma^{\text{Hecke}}\) is an equivalence of abelian and derived categories. The main theorem of the current paper proves the conjecture for a group \(G=GL_2\) as well as simple groups of rank 1. Let us give an idea of the proof. \begin{itemize} \item Frenkel and Gaitsgory proved that \(\Gamma^{\text{Hecke}}\) is fully faithful and that \(\Gamma^{\text{Hecke}}\) is an equivalence on \(I^0\)-equivariant objects, where \(I^0\) is the radical of the Iwahori subgroup \(I\); \item Earlier results of Raskin's ``W-algebras and Whittaker categories and Frenkel-Gaitsgory-Vilonen's ``Whittaker patterns in the geometry of moduli spaces of bundles on curves showed that \(\Gamma^{\text{Hecke}}\) is an equivalence on Whittaker objects; \item The new result of the current paper is that for \(G=PGL_2\), any \(G(K)\)-category \(C\) with a level \(\kappa\) is generated under \(G(K)\) by the \(I^0\)-equivariant objects and the Whittaker objects. \end{itemize} Note that the first two points work for any reductive group \(G\). Then by observing that the essential image of \(\Gamma^{\text{Hecke}}\) is a \(G(K)\)-category, one can prove the result. With some more work, this can also be used to prove t-exactness. Later results of \textit{D. Yang} and \textit{S. Raskin} prove the conjecture for a general reductive group \(G\) [``Affine Beilinson-Bernstein localization at the critical level, Preprint, \url{arXiv:2203.13885}], using categorical Moy-Prasad theory developed by \textit{D. Yang} [``Categorical Moy-Prasad theory Yang, Preprint, \url{arXiv:2104.12917}] as a main new ingredient.



Cites work









This page was built for publication: Affine Beilinson-Bernstein localization at the critical level for \(\mathrm{GL}_2\)

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2078804)