betaFeedback

WorksMathematics

Sections of the Kottwitz homomorphism and affine pinnings

Abstract

Let GG be a connected reductive group split over Z\mathbb{Z}. The Kottwitz homomorphism identifies the group of connected components of the loop group of GG with the group ΩΩ of length zero elements of the extended affine Weyl group. We show that this homomorphism always has a homomorphic section, and that the section can be chosen inside the normalizer of TT in G(Z[t,t−1])G(\mathbb{Z}[t,t^{-1}]). We then ask whether the section can be chosen to preserve an affine pinning. This is possible if no simple factor of GG is of type B2mB_{2m} or C2mC_{2m}. It is impossible for (SL2×Sp2n)/μ2(SL_2\times Sp_{2n})/μ_2 and (SL2×Spin2n+1)/μ2(SL_2\times Spin_{2n+1})/μ_2 when nn is even. We show that the obstructions for GG and for its Langlands dual group are dual to each other, and we compute them by a lattice computation with a Coxeter element. As an application, we show that the component group of the centralizer of any semisimple element lifts to the centralizer. Under the same condition on the simple factors, the lift can be chosen to preserve a pinning of the neutral component.

Introduction

Loop groups and their components

Let kk be a field. For a linear algebraic group GG over kk, the loop group LGLG is the functor R↦G(R((t)))R \mapsto G(R((t))) on kk-algebras. It is a group ind-scheme with kk-points G(k((t)))G(k((t))). A reader who is not familiar with ind-schemes may simply think of the abstract group G(k((t)))G(k((t))). Loop groups are the function-field analogues of pp-adic groups. Their quotients by parahoric subgroups, such as the affine Grassmannian and the affine flag variety, are basic objects in geometric representation theory [17, 3 Zhu17].

The name comes from topology. Over C\mathbb{C}, the group G(C[t,t−1])G(\mathbb{C}[t,t^{-1}]) consists of algebraic maps C×→G(C)\mathbb{C}^{\times} \to G(\mathbb{C}), and these restrict to loops in G(C)G(\mathbb{C}). The components of the space of loops in a connected Lie group HH are indexed by π1(H)\pi_1(H). Algebraic loop groups behave in the same way. Let GG be connected reductive and split, with split maximal torus TT and coroot lattice Q∨⊂X∗(T)Q^\vee\subset X_*(T). Then the components of LGLG are indexed by the algebraic fundamental group

π1(G)=X∗(T)/Q∨\pi_1(G) = X_*(T)/Q^\vee

[18 Theorem 5.1]. So LGLG is disconnected unless GG is semisimple and simply connected.

Here are the basic examples. For G=GLnG = \mathrm{GL}_n, the map g↦val⁡(det⁡g)g \mapsto\operatorname{val}(\det g) is a surjective homomorphism GLn(k((t)))→Z=π1(GLn)\mathrm{GL}_n(k((t))) \to\mathbb{Z} = \pi_1(\mathrm{GL}_n). Its fibres are the components of LGLnLGL_n. For SLnSL_n the loop group is connected. For PGLnPGL_n we have PGLn(k((t)))=GLn(k((t)))/k((t))×PGL_n(k((t))) = GL_n(k((t)))/k((t))^\times, and val⁡∘det⁡\operatorname{val} \circ\det induces a surjection onto Z/nZ=π1(PGLn)\mathbb{Z}/n\mathbb{Z} = \pi_1(PGL_n).

In these examples, the projection onto the group of components has a homomorphic section. For GLnGL_n we can take m↦diag⁡(tm,1,…,1)m \mapsto\operatorname{diag}(t^m,1,\ldots,1). For PGLnPGL_n, consider

γ=(0Id⁡n−1t0)∈GLn(k((t))).\gamma= \begin{pmatrix} 0 & \operatorname{Id}_{n-1}\\ t & 0 \end{pmatrix} \in GL_n(k((t))).

Then γn=t⋅Id⁡n\gamma^n = t \cdot\operatorname{Id}_n is a scalar matrix, and val⁡(det⁡γ)=1\operatorname{val}(\det\gamma) = 1. So the image of γ\gamma in PGLn(k((t)))PGL_n(k((t))) has order nn, and j↦γjj \mapsto\gamma^j is a homomorphic section. The element γ\gamma normalizes the standard Iwahori subgroup. It rotates the affine Dynkin diagram of type An−1(1)A_{n-1}^{(1)}, which is an nn-cycle.

The problem

Does such a section exist for every split reductive group? A natural strategy is to lift each component to a monomial element wtμw t^\mu. Here w˙\dot{w} lies in the normalizer NG(T)N_G(T), and tμ=μ(t)t^\mu= \mu(t) is a cocharacter evaluated at tt. The matrix γ\gamma is of this form.

The difficulty is that elements of the Weyl group W=NG(T)/TW = N_G(T)/T often cannot be lifted to NG(T)N_G(T) without increasing their order. For example, every element of NSL2(T)N_{SL_2}(T) over the non-trivial Weyl element has the form (0a−a−10)\begin{pmatrix} 0 & a \\ -a^{-1} & 0 \end{pmatrix}, and its square is −Id⁡-\operatorname{Id}. Tits [22] constructed canonical lifts nw∈NG(T)n_w \in N_G(T) of all w∈Ww \in W. They satisfy the braid relations, but w↦nww \mapsto n_w is not a homomorphism.

For split almost simple groups over pp-adic fields, Adrian constructed a homomorphic section from Tits’ lifts [1 Theorem 3.5]. A new problem appears when several simple factors are glued together through the center. Consider

G=(SL2×Sp4)/μ2,G = (SL_2 \times Sp_4)/\mu_2,

where μ2\mu_2 is embedded diagonally into the center. Then π1(G)≅Z/2Z\pi_1(G) \cong\mathbb{Z}/2\mathbb{Z}. A monomial lift of the generator is the image in GG of a pair (x,y)(x,y), times a cocharacter evaluated at tt. Here x∈SL2x \in SL_2 lies over the non-trivial Weyl element, and y∈Sp4y \in Sp_4 lies over a certain involution ww. The cocharacter cancels in the square, so the square of the lift is the image of (x2,y2)(x^2,y^2). Hence the lift has order 22 if and only if

y2=x2=−Id⁡.y^2 = x^2 = -\operatorname{Id}.

But Tits’ lift satisfies nw2=Id⁡n_w^2 = \operatorname{Id} (Example 2.5). So nwn_w must be replaced by another lift. Such a lift exists, but it does not respect the natural symmetries of the loop group (Theorem B).

Main results

Let GG be a connected reductive group split over Z\mathbb{Z}, with split maximal torus TT and Weyl group WW. Let W~=X∗(T)⋊W\widetilde{W} = X_*(T) \rtimes W be the extended affine Weyl group. We write its elements as wtμw t^\mu. Let Ω⊂W~\Omega\subset\widetilde{W} be the subgroup of elements of length zero. It is the stabilizer of the fundamental alcove, and it acts on the affine Dynkin diagram. The map wtμ↦[μ]w t^\mu\mapsto[\mu] gives an isomorphism Ω≅π1(G)\Omega\cong\pi_1(G) [9 Lemma 14].

For a field kk, the projection of LGkLG_k onto its group of components is the Kottwitz homomorphism κG\kappa_G [12 §7], [18 §5]. We obtain an exact sequence

1⟶LGk∘⟶LGk→κGΩ⟶1,(1)1 \longrightarrow LG_k^\circ\longrightarrow LG_k \xrightarrow{\kappa_G} \Omega\longrightarrow1, \tag*{(1)}

where LGk∘LG_k^\circ is the neutral component, and κG(tμ)=[μ]\kappa_G(t^\mu) = [\mu].

The monomial elements are defined over Z\mathbb{Z}. Every n∈NG(T)(Z[t,t−1])n \in N_G(T)(\mathbb{Z}[t,t^{-1}]) can be written as n=τnwtμn=\tau n_wt^\mu with τ∈T(Z)\tau\in T(\mathbb{Z}). The map n↦wtμn\mapsto wt^\mu is a surjective homomorphism from NG(T)(Z[t,t−1])N_G(T)(\mathbb{Z}[t,t^{-1}]) onto W~\widetilde{W}, with kernel T(Z)≅(Z/2Z)rk⁡TT(\mathbb{Z})\cong(\mathbb{Z}/2\mathbb{Z})^{\operatorname{rk}T} (Lemma 2.7). Our first result says that this extension splits over Ω\Omega.

Theorem A. Let GG be a connected reductive group split over Z\mathbb{Z}. There is a group homomorphism

ψ:Ω→NG(T)(Z[t,t−1])\psi:\Omega\to N_G(T)(\mathbb{Z}[t,t^{-1}])

such that ψ(ω)\psi(\omega) maps to ω\omega in W~\widetilde{W} for every ω∈Ω\omega\in\Omega. Consequently, for every field kk, the composite

Ω→ψNG(T)(Z[t,t−1])→LG(k)\Omega\xrightarrow{\psi}N_G(T)(\mathbb{Z}[t,t^{-1}])\to LG(k)

is a homomorphic section of κG\kappa_G. In particular, LGk≅LGk∘⋊ΩLG_k\cong LG_k^\circ\rtimes\Omega.

This is Theorem 4.4. Replacing tt by a uniformizer gives a homomorphic section of the Kottwitz homomorphism G(F)→ΩG(F)\to\Omega for every non-archimedean local field FF (Corollary 4.5). This extends the results of Adrian from almost simple groups to all split reductive groups.

Pinned splittings. A pinning of GG consists of a Borel subgroup B⊃TB\supset T and a non-zero root vector for each simple root. Automorphisms of GG that preserve a pinning correspond to automorphisms of the Dynkin diagram. The neutral component LG∘LG^\circ has an analogous structure, an affine pinning. It consists of an Iwahori subgroup I⊃TI\supset T and a root vector for each simple affine root (Definition 2.10).

We say that GG is pinnable over kk if {(1.1)} has a splitting whose image preserves an affine pinning. Suppose GG is semisimple, and let Ω~⊂LG(k)\widetilde{\Omega}\subset LG(k) be the stabilizer of an affine pinning. Then there is an exact sequence (Proposition 2.12 and Corollary 5.4; see also [6 Proposition 3.14])

1→Z(G)(k)→Ω~→Ω→1(2)1\to Z(G)(k)\to\widetilde{\Omega}\to\Omega\to1 \tag*{(2)}

and GG is pinnable if and only if {(1.2)} splits. When GG is adjoint, Z(G)Z(G) is trivial, so GG is pinnable.

Theorem B. Let GG be a connected reductive group split over Z\mathbb{Z}, and let kk be a field.

  1. Suppose that GG is almost simple, or that no simple factor of GG is of type B2mB_{2m} or C2mC_{2m}. Then GG is pinnable over kk. Moreover, the section ψ\psi in Theorem A can be chosen to preserve an affine pinning defined over Z\mathbb{Z}.

  1. Suppose char⁡k≠2\operatorname{char}k\ne2, and let n≥2n\ge2 be even. Let GG be (SL⁡2×Sp⁡2n)/μ2(\operatorname{SL}_2\times\operatorname{Sp}_{2n})/\mu_2 or (SL⁡2×Spin⁡2n+1)/μ2(\operatorname{SL}_2\times\operatorname{Spin}_{2n+1})/\mu_2, with μ2\mu_2 embedded diagonally into the center. Then Ω≅Z(G)(k)≅Z/2Z\Omega\cong Z(G)(k)\cong\mathbb{Z}/2\mathbb{Z}, but every lift of the generator of Ω\Omega that preserves an affine pinning has order 44. In particular, GG is not pinnable over kk.

See Theorem 5.5 and Proposition 5.6. So {(1.1)} always splits, but a pinned splitting may not exist. The two groups in (2) are Langlands dual to each other. This is an instance of Theorem C below.

References

  1. [1][Adr18] M. Adrian, A remark on the Kottwitz homomorphism, Manuscripta Math. 155 (2018), no. 1–2, 1–14.arxiv.org/abs/1608.03601
  2. [2][AH17] J. Adams and X. He, Lifting of elements of Weyl groups, J. Algebra 485 (2017), 142–165.arxiv.org/abs/1608.00510
  3. [3][Bez16] R. Bezrukavnikov, On two geometric realizations of an affine Hecke algebra, Publ. Math. Inst. Hautes Études Sci. 123 (2016), 1–67.arxiv.org/abs/1209.0403
  4. [4][Bou02] N. Bourbaki, Lie groups and Lie algebras. Chapters 4–6, Elements of Mathematics, Springer-Verlag, Berlin, 2002.
  5. [5][Con14] B. Conrad, Reductive group schemes, Autour des schémas en groupes. Vol. I, Panor. Synthèses, vol. 42/43, Soc. Math. France, Paris, 2014, pp. 93–444.
  6. [6][DLYZ25] G. Dhillon, Y. W. Li, Z. Yun, and X. Zhu, Endoscopy for metaplectic affine Hecke categories, 2025, arXiv:2507.16667.arxiv.org/abs/2507.16667
  7. [7][GH24] R. Ganapathy and X. He, Tits groups of Iwahori-Weyl groups and presentations of Hecke algebras, Transform. Groups 29 (2024), no. 4, 1419–1457.arxiv.org/abs/2107.01768
  8. [8]T. J. Haines, J. Lourenço, and T. Richarz, On the normality of Schubert varieties: remaining cases in positive characteristic, Ann. Sci. Éc. Norm. Supér. (4) 57 (2024), no. 3, 895–959.arxiv.org/abs/1806.11001
  9. [9]T. Haines and M. Rapoport, On parahoric subgroups, Adv. Math. 219 (2008), no. 1, 188–198.arxiv.org/abs/0804.3788
  10. [10]N. Iwahori and H. Matsumoto, On some Bruhat decomposition and the structure of the Hecke rings of p-adic Chevalley groups, Publ. Math. Inst. Hautes Études Sci. 25 (1965), 5–48.DOI
  11. [11]B. Kostant, The principal three-dimensional subgroup and the Betti numbers of a complex simple Lie group, Amer. J. Math. 81 (1959), 973–1032.
  12. [12]R. E. Kottwitz, Isocrystals with additional structure. II, Compositio Math. 109 (1997), no. 3, 255–339.DOI
  13. [13]Y. W. Li, Endoscopy for affine Hecke categories, Selecta Math. (N.S.) 30 (2024), no. 5, Paper No. 93.DOI
  14. [14]G. Lusztig, Lifting involutions in a Weyl group to the torus normalizer, Represent. Theory 22 (2018), 27–44.arxiv.org/abs/1709.08589
  15. [15]G. Lusztig and Z. Yun, Endoscopy for Hecke categories, character sheaves and representations, Forum Math. Pi 8 (2020), e12.
  16. [17]I. Mirković and K. Vilonen, Geometric Langlands duality and representations of algebraic groups over commutative rings, Ann. of Math. (2) 166 (2007), no. 1, 95–143.arxiv.org/abs/math/0401222
  17. [18]G. Pappas and M. Rapoport, Twisted loop groups and their affine flag varieties, Adv. Math. 219 (2008), no. 1, 118–198, With an appendix by T. Haines and M. Rapoport.arxiv.org/abs/math/0607130
  18. [19]Schémas en groupes. III: Structure des schémas en groupes réductifs, Lecture Notes in Math., vol. 153, Springer-Verlag, Berlin, 1970, Séminaire de Géométrie Algébrique du Bois Marie 1962/64 (SGA 3), dirigé par M. Demazure et A. Grothendieck.
  19. [20]T. A. Springer, Linear algebraic groups, second ed., Progress in Mathematics, vol. 9, Birkhäuser, Boston, 1998.
  20. [21]T. A. Springer and R. Steinberg, Conjugacy classes, Seminar on Algebraic Groups and Related Finite Groups (Princeton, 1968/69), Lecture Notes in Math., vol. 131, Springer, Berlin, 1970, pp. 167–266.DOI
  21. [22]J. Tits, Normalisateurs de tores. I. Groupes de Coxeter étendus, J. Algebra 4 (1966), 96–116.DOI

Paper details

vlnc:W_prku3qgf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents