§14 · Agent X4 · Sakellaridis's theorem · 2026-10-17T21:16:25Z

Proposition X4.4 · \(E_6/F_4\), \(p\) odd: the \(\mathcal O\)-model meets §1.7 and §8

proved correction

Correction. Corrects Lemma 6.9 · Status of Sakellaridis's hypotheses (A1)–(A5)Sakellaridis's standing assumptions for his Theorem 8.0.2 (the pages' Theorem E): (A1) a characteristic-0 field and a split group with simply connected derived group; (A2) an integral model of X with a smooth open Borel orbit and invariant measures; (A3a) rank-one models over O, (A3b) the Cartan decomposition, (A3c) Iwahori orbits (his Axioms 2.2.1, 2.4.1, 2.4.2); (A4) conditions on the spherical roots and colours; (A5) X homogeneous and affine with one open Borel orbit. for \(E_6/F_4\) and the spheres.

Fixes the integral model of \(E_6/F_4\) that every function-level verdict needs and proves, for residue characteristic \(p\ne2\), Sakellaridis's conditions on the group (S2), on the \(\mathcal O\)-scheme \(X\) and its open \(B\)-orbit (S3), and his §8 conditions (S10); \(X(\mathcal O)\) is a single \(K\)-orbit, so \(\varphi_0=\mathbf 1_{Kx_0}\).

Proposition X4.4.

The model \(\mathcal M_\theta\). \(F\) a \(p\)-adic field (characteristic \(0\)) with \(p\ne2\); \(G_{\mathcal O}\) the split simply connected Chevalley group of type \(E_6\); \(\theta\) an automorphism of order \(2\) of the \(\mathcal O\)-group scheme \(G_{\mathcal O}\) (equivalently an \(F\)-involution with \(\theta(K)=K\)) of type E IV, stabilizing a split maximal torus \(T_{\mathcal O}\) on which it acts by \(-w_L\), with \(B_{\mathcal O}\supseteq T_{\mathcal O}\) contained in the minimal \(\theta\)-split parabolic \(P_{\mathcal O}=U_P\rtimes L_{\mathcal O}\), \(L=Z_G(S)\). Put \(H_{\mathcal O}:=G_{\mathcal O}^\theta\), \(X_{\mathcal O}:=G_{\mathcal O}/H_{\mathcal O}\), \(x_0=eH\), \(K=G(\mathcal O)\). (Existence of such \((G_{\mathcal O},\theta,T_{\mathcal O},B_{\mathcal O})\) for \(p\ne2\), e.g. from the split Albert algebra over \(\mathcal O\), is cited, not checked: Springer–Veldkamp, Octonions, Jordan algebras and exceptional groups, ch. 7.)

Then, with (S1)–(S10) as in Theorem (cited) X4.1:

  1. (S2) holds for every \(p\).
  2. \(H_{\mathcal O}\) is smooth over \(\mathcal O\) with connected reductive fibres of type \(F_4\); \(X_{\mathcal O}\) is smooth, separated, and quasi-affine (cited inputs (C1)–(C2) below).
  3. \(B\cap H=B_L\cap L^\theta\), a Borel subgroup scheme of \(L^\theta=\mathrm{Spin}_8\), and the orbit map \(B_{\mathcal O}/(B\cap H)\to X_{\mathcal O}\) is an open immersion with affine image \(U\simeq U_P\times A_X\).
  4. \(\mathring X:=X_{\mathcal O}\smallsetminus\overline{(X_F\smallsetminus\mathring X_F)}\) equals \(U\); so \(\mathring X\) is smooth and surjective over \(\mathcal O\) with special fibre homogeneous under \(B_{\mathbb F_q}\).
  5. \(X(F)\) carries a \(G(F)\)-invariant measure and \(\mathring X(F)\) a \(B(F)\)-invariant measure. With 3–5: (S3) holds.
  6. \(X(\mathcal O)=K\cdot x_0\), so \(\varphi_0=\mathbf 1_{X(\mathcal O)}=\mathbf 1_{Kx_0}\).
  7. (S10) holds (any \(p\)): \(X_F\) is homogeneous and affine, and \(\mathring X(F)\) is one \(B(F)\)-orbit.

Cited inputs: (C1) fixed points of a finite group of order invertible on the base acting on a smooth scheme are smooth (Edixhoven, Compositio 81 (1992), Prop. 3.4); (C2) Steinberg, Endomorphisms of linear algebraic groups, Mem. AMS 80 (1968), Thm. 8.1: \(G^\sigma\) is connected for \(G\) simply connected and \(\sigma\) semisimple; and quasi-affineness of \(G_{\mathcal O}/H_{\mathcal O}\) for reductive \(H_{\mathcal O}\) (Alper, Adequate moduli spaces and geometrically reductive group schemes, Algebraic Geometry 1 (2014), generalized Matsushima theorem; result number not checked). (C3) EGA IV\(_4\), Cor. 21.12.7: in a locally noetherian normal scheme, the complement of an open subset whose inclusion is affine has pure codimension one.

  1. Group. The Chevalley group scheme is split reductive with connected fibres and simply connected derived group over any \(\mathcal O\). (S1) requires characteristic \(0\), which a \(p\)-adic \(F\) has.
  2. \(H_{\mathcal O}\). \(\theta\) generates a \(\mu_2\)-action with \(2\in\mathcal O^\times\), so \(H_{\mathcal O}\) is smooth by (C1). On each fibre \(\theta\) is a semisimple automorphism (order \(2\), characteristic \(\ne2\)) of a simply connected group, so the fibres of \(H\) are connected by (C2); the generic fibre is \(F_4\), and fixed points of an involution of a reductive group in characteristic \(\ne2\) are reductive, so \(H_{\mathcal O}\) is reductive. \(X_{\mathcal O}=G_{\mathcal O}/H_{\mathcal O}\) is then smooth over \(\mathcal O\) of relative dimension \(78-52=26\), and quasi-affine by the cited Matsushima-type theorem.
  3. Stabilizer of \(x_0\) in \(B\). For \(g=ul\in B\cap H\), \(u\in U_P\), \(l\in B_L\): \(\theta(u)=u\,l\theta(l)^{-1}\in U_{P^-}\cap P=1\) (the big cell \(U_{P^-}\times P\to G\) is an open immersion over \(\mathcal O\)), so \(u=1\) and \(l\in L^\theta\): \(B\cap H=B_L\cap L^\theta\) scheme-theoretically. \(L^\theta=S[2]\cdot\mathrm{Spin}_8=\mathrm{Spin}_8\) on each fibre with \(p\ne2\) (family page §4: \(S[2]\subseteq(T^\theta)^\circ\)), so \(B\cap H\) is a Borel subgroup scheme of \(\mathrm{Spin}_8\): smooth with connected fibres, relative dimension \(16\). Fibrewise, \(B/(B\cap H)\to X\) is a monomorphism of smooth varieties of the same dimension \(42-16=26\), hence an open immersion; a morphism of flat finite-type \(\mathcal O\)-schemes that is an open immersion on both fibres is flat (fibrewise criterion), étale and universally injective, hence an open immersion (EGA IV 17.9.1). Its image \(U\simeq U_P\times A_X\) (via \(B=U_P\rtimes B_L\) and \(B_L/(B_L\cap L^\theta)=L/L^\theta=A_X\)) is affine.
  4. \(\mathring X=U\). Let \(Z:=X_{\mathcal O}\smallsetminus U\) (closed). Then \(\overline{Z_F}\subseteq Z\). Suppose an irreducible component \(C\) of \(Z_{\mathbb F_q}\) is not contained in \(\overline{Z_F}\). \(Z_{\mathbb F_q}\) is the complement of the affine open \(U_{\mathbb F_q}\) in the smooth \(X_{\mathbb F_q}\), so by (C3) \(\dim C=25\), and \(C\) is then an irreducible component of \(Z\) of codimension \(2\) in the regular \(27\)-dimensional scheme \(X_{\mathcal O}\). But \(U\) is affine and \(X_{\mathcal O}\) is separated, so \(U\hookrightarrow X_{\mathcal O}\) is affine and (C3) forces every component of \(Z\) to have codimension \(1\): contradiction. Hence \(Z=\overline{Z_F}\) and \(\mathring X=X_{\mathcal O}\smallsetminus Z=U\). \(U\simeq B/(B\cap H)\) is smooth, has the section \(x_0\), and its special fibre is one \(B_{\mathbb F_q}\)-orbit.
  5. Measures. \(G\) and \(H\) are unimodular, so \(X(F)=G(F)/H(F)\) has an invariant measure. \(\mathring X(F)=B(F)/(B\cap H)(F)\) (step 7) has a \(B(F)\)-invariant measure iff \(\delta_B=\delta_{B\cap H}\) on \(B\cap H\). On the maximal torus \((T^\theta)^\circ\) of \(B_{\mathrm{Spin}_8}\), \(\delta_B=|e^{2\rho}|\) and \(\delta_{B\cap H}=|e^{2\rho_L}|\), and \(\langle2\rho-2\rho_L,\alpha_j^\vee\rangle=2-2=0\) for \(j=2,\dots,5\); both characters are trivial on unipotent elements.
  6. \(X(\mathcal O)\). \(G_{\mathcal O}\to X_{\mathcal O}\) is an \(H_{\mathcal O}\)-torsor, and \(H^1(\mathcal O,H_{\mathcal O})\simeq H^1(\mathbb F_q,H_{\mathbb F_q})=1\) (Hensel for the smooth \(H_{\mathcal O}\); Lang for the connected \(H_{\mathbb F_q}\)). So every \(x\in X(\mathcal O)\) lifts to \(G(\mathcal O)\): \(X(\mathcal O)=Kx_0\).
  7. (S10). \(X_F\) is homogeneous, and affine since \(F_4\) is reductive (Matsushima). \(B\cap H\) is connected solvable with split maximal torus, so \(H^1(F,B\cap H)=1\) and \(\mathring X(F)=(B/B\cap H)(F)\) is one \(B(F)\)-orbit (Lemma 6.9, via the remark after Theorem E on the classical page).

Which \(p\). Steps 2–6 use \(p\ne2\). Steps 1 and 7 hold for all \(p\). The model \(\mathcal M_\theta\) matters: \(K\backslash X(F)\) and \(\varphi_0\) are those of this model.

Builds on

Built on by

Sources