§1 · Agent 1 · Direct rerun · 2026-10-18T04:00:32Z

Proposition 1.6 · \(F_wS_\chi=b_wS_{w\chi}\) and the sum formula (6.2) over \(\mathbb F_q((t))\)

proved audit: correct ★ key step

Second audit: correct. (a) re-derived: wD=D iff w is in N_W(a) and w rho_L=rho_L (Q Phi_L injects into X^(T^theta)Q) iff w is minimal in wW_L; for E6 I found exactly six such elements. (b) I checked it with my own rank-one Iwasawa decompositions (script rank1.py): (4.8); F_s^2=1 with the (4.9) factor taken at the source; F_w^=F{w^{-1}}; and K_beta-invariance of the pair {w, w s_beta} (coefficient ratio -e^{w beta-check}(chi)). This reduces Proposition 4.2.1 in I(chi) to SL2, without [BK98] or [Ca80]. (c) and (d) re-derived: uniqueness at w chi in D0D0 · Orbit geometry over F̄_q, p ≠ 2 (Chen–Yi, Prop. 17, Thm 18): orbits indexed by Λ_S^+, closures, the support bound and the stabilizers.' gives rational, nonzero b_w with the cocycle rule, and Lemma 3.9.1 needs only xJ in xB(O). Reindexing v=w^{-1} gives B_v=c(v)(chi) b_v(chi^{-1}), and by adjunction b_v(chi^{-1}) is Sakellaridis's b_v, so the formula is exactly (6.1)-(6.2) with W_X'. All of this is valid over any non-archimedean local field with psi of conductor O.

Derives Sakellaridis's sum formula for the Hecke eigenfunction Ω_χ on A_X^+ in characteristic p from the generic uniqueness of Proposition 1.5, normalized intertwining operators and the SL_2 Fourier identity; the summation group is W_X′ = N_W(a)∩[W/W_L], defined without Knop's action.

Proposition 1.6.

Setting of Proposition 1.5, with (A2)(A2) Integral model: X quasi-affine over O, its open Borel orbit smooth and surjective over O with B-homogeneous special fibre; a G-eigenmeasure on X and a B-invariant measure on the open orbit (§1.7). and (A3c)(A3c) Iwahori orbits (Sakellaridis's Axiom 2.4.2): xJ ⊆ xB(O) for every X-antidominant torus point x.: \(xJ\subseteq xB(O)\) for \(x\in A_X^+\), where \(A_X^+\) is the set of points with antidominant label. Let \(D=\delta(X)^{1/2}A_X^*\), \(D'=\delta(X)^{-1/2}A_X^*=w_LD\), and \(\mathfrak a=X_*(S)_{\mathbb Q}\). Define $$W_X':=\{w\in W:\ w(D)=D\}.$$ (a) \(W_X'=N_W(\mathfrak a)\cap[W/W_L]\). It is a group mapping isomorphically onto \(N_W(\mathfrak a)/W_L\), it commutes with \(w_L\), and it fixes \(\delta(X)\).

(b) Let \(F_w:=\prod_{\check\alpha>0,\,w\check\alpha<0}\frac{1-e^{\check\alpha}}{1-q^{-1}e^{-\check\alpha}}(\chi)\,T_w:I(\chi)\to I(w\chi)\) (eq. (4.9)). Then \(F_{w_1w_2}=F_{w_1}F_{w_2}\) for all \(w_1,w_2\), and \(F_w^*=F_{w^{-1}}\) for the pairing of \(I(\chi)\) with \(I(\chi^{-1})\). The vector \(\Phi_{K,\chi}:=\operatorname{vol}(J)^{-1}\sum_{w\in W}\prod_{\alpha>0,w^{-1}\alpha<0}(-e^{\check\alpha}(\chi))\,F_w\varphi_{J,w^{-1}\chi}\) is \(K\)-invariant (Proposition 4.2.1).

(c) For generic \(\chi\in D'\): if \(w\in W_X'\) then \(F_w\circ S_\chi=b_w(\chi)S_{w\chi}\), with \(b_w\) rational on \(D'\) and \(b_{w_1w_2}(\chi)=b_{w_1}(w_2\chi)b_{w_2}(\chi)\). If \(w\notin W_X'\), the functional \(\mathrm{ev}_1\circ F_wS_\chi\) vanishes on \(C_c^\infty(X̊(F))\).

(d) (Theorem 4.2.2 / eq. (6.2), formula part) Let \(\Delta_\chi\) be the adjoint of \(S_{\chi^{-1}}\) and \(\Omega_\chi:=\Delta_\chi(\Phi_{K,\chi})\) for \(\chi\in D\). For \(x\in A_X^+\) and generic \(\chi=\delta(X)^{1/2}\tilde\chi\), $$\Omega_\chi(x)=\delta_{P(X)}^{-1/2}(x)\sum_{w\in W_X'}B_w(\tilde\chi)\,{}^w\tilde\chi(x),\qquad B_w(\tilde\chi)=\prod_{\check\alpha>0,\,w\check\alpha<0}(-e^{\check\alpha}(\chi))\,b_w,$$ with \(b_w\) transported from \(D'\) to \(D\) through the adjoint (\(F_w^*=F_{w^{-1}}\)). \(\Omega_\chi\) is a \(K\)-invariant Hecke eigenfunction with eigencharacter \(\chi\).

This replaces Theorem 3.7.1/Corollary 3.7.2 of [Sa08], the Braverman–Kazhdan formalism [BK98] and [Ca80, Thm 3.1] on the route to (4.2). All arguments are valid over any non-archimedean local field \(F\) once \(\psi\) has conductor \(O\) and \(dx(O)=1\).

Proof.

  1. (a). \(w(D)=D\) forces \(w\) to normalize the subtorus \(A_X^*\), i.e. \(w\in N_W(\mathfrak a)\). Such \(w\) permutes \(\Phi_L=\Phi\cap\mathfrak a^\perp\). Also \(w(\delta(X)^{1/2})\delta(X)^{-1/2}=|e^{w\rho_L-\rho_L}|\), and \(w\rho_L-\rho_L\in\mathbb Q\Phi_L\), which restricts injectively to \(T^\theta\). So \(w(\delta(X)^{1/2})\in\delta(X)^{1/2}A_X^*\) iff \(w\rho_L=\rho_L\) iff \(w\Phi_L^+=\Phi_L^+\) iff \(w\) is the shortest element of \(wW_L\).

The pointwise stabilizer of \(\mathfrak a\) is the reflection subgroup generated by the roots orthogonal to \(\mathfrak a\), namely \(W_L\) (Chevalley–Steinberg). So \(W_X'\cong N_W(\mathfrak a)/W_L\). Elements of \(W_X'\) normalize \(\Phi_L^+\), hence commute with \(w_L\) and fix \(\rho_L\). For \(G\) split and \(\operatorname{char}\ne2\), \(N_W(\mathfrak a)/Z_W(\mathfrak a)\) is the restricted Weyl group (Richardson 1982; cited). For \(M_\theta\) it is computed directly in Computation 1.7.

  1. (b), rank one. The function \(u\mapsto\varphi_{K,\chi}(\tilde w_\alpha u_\alpha(x))\) equals 1 for \(|x|\le1\). For \(|x|=q^n\ge q\) the \(SL_2\) identity \(\tilde w u(x)\in u'\check\alpha(x^{-1})K\) gives the value \(e^{-n\check\alpha}(\chi\delta^{1/2})\). Summing with \(\operatorname{vol}\{|x|=q^n\}=q^n(1-q^{-1})\) gives \(T_\alpha\varphi_K=c_\alpha(\chi)\varphi_K\) with \(c_\alpha=\frac{1-q^{-1}e^{\check\alpha}}{1-e^{\check\alpha}}\), the rank-one case of (4.8).

\(T_{w_1w_2}=T_{w_1}T_{w_2}\) when lengths add (Fubini). For generic \(\chi\), \(\operatorname{End}_G I(\chi)=\mathbb C\) (Bruhat filtration / geometric lemma of Bernstein–Zelevinsky, valid for \(\ell\)-groups). Hence \(T_\alpha^2=c_\alpha(\chi)c_\alpha(w_\alpha\chi)\), and with the normalization (4.9), \(F_\alpha^2=1\). So the \(F_w\) satisfy the relations of \(W\), and \(F_{w_1w_2}=F_{w_1}F_{w_2}\) always. \(T_w^*=T_{w^{-1}}\) and the symmetric normalizing factor give \(F_w^*=F_{w^{-1}}\). The \(K\)-invariance of \(\Phi_{K,\chi}\) is Sakellaridis's proof of Proposition 4.2.1 projected to \(I(\chi)\). It uses only \(F_{ww_\beta}=F_wF_{w_\beta}\) and the \(SL_2\) identity \(\mathcal F\mathbf 1_{\mathfrak p\times O}=q^{-1}\mathbf 1_{O\times\mathfrak p^{-1}}\) for the symplectic Fourier transform (3.10) on \(F^2\). That identity is a one-line computation with \(\psi\) of conductor \(O\).

  1. (c), \(w\in W_X'\). \(F_wS_\chi\in\operatorname{Hom}_G(C_c^\infty(X(F)),I(w\chi))\) and \(w\chi\in D'\) (since \(W_X'\) commutes with \(w_L\), it also stabilizes \(D'\)). By Proposition 1.5(b), for generic \(\chi\) this space is the line spanned by \(S_{w\chi}\). By Proposition 1.5(c), \(b_w\) is rational. The cocycle relation follows from \(F_{w_1w_2}=F_{w_1}F_{w_2}\).
  1. (c), \(w\notin W_X'\). The restriction of \(\mathrm{ev}_1\circ F_wS_\chi\) to \(C_c^\infty(X̊(F))=C_c^\infty(B(F)/B_{x_0}(F))\) is a \((B,w\chi)\)-equivariant functional. It is non-zero only if \(w\chi\) satisfies (3.3) for \(x_0\), i.e. \(w\chi\in D'\). Since \(w(D')\neq D'\) and both are irreducible of the same dimension, \(w\chi\notin D'\) for generic \(\chi\).
  1. (d). For \(f\in I(\chi)^J\) and \(x\in A_X^+\), \(J\)-invariance gives \(\Delta_\chi(f)(x)=\langle f,S_{\chi^{-1}}(\mathbf 1_{xJ}|\omega_X|)\rangle/|\omega_X|(xJ)\). By (A3c)(A3c) Iwahori orbits (Sakellaridis's Axiom 2.4.2): xJ ⊆ xB(O) for every X-antidominant torus point x., \(\mathbf 1_{xJ}\in C_c^\infty(X̊(F))\). By steps 3–4 and \(F_w^*=F_{w^{-1}}\), the term of \(w\) in \(\Delta_\chi(\Phi_{K,\chi})(x)\) is

\(\langle\varphi_{J},F_{w^{-1}}S_{\chi^{-1}}(\mathbf 1_{xJ})\rangle\). This is 0 unless \(w^{-1}\in W_X'\), in which case it equals \(b_{w^{-1}}(\chi^{-1})\langle\varphi_J,S_{w^{-1}\chi^{-1}}(\mathbf 1_{xJ})\rangle\). Lemma 3.9.1 (whose proof uses only \(xJ\subseteq xB(O)\)) evaluates the last pairing as \(\operatorname{vol}(J)\,|\omega_X|(xJ)\) times the character \({}^{w}\tilde\chi\delta^{-1/2}\) at \(x\). Collecting the factors \(\prod(-e^{\check\alpha})\) as in the passage from Proposition 4.2.1 to Theorem 4.2.2 and to (6.1)–(6.2) gives the formula. \(\Phi_{K,\chi}\) is \(K\)-invariant and \(\Delta_\chi\) is a \(G\)-morphism, so \(\Omega_\chi\) is a \(K\)-invariant eigenfunction. It is non-zero because the term \(w=1\) contributes the character \(\tilde\chi\delta_{P(X)}^{-1/2}\), which is linearly independent of the others for generic \(\chi\) (the remark at the end of §3.10).

What is not used. Knop's action, the orbits \(wX̊\ne X̊\), Brion's paths and Proposition 3.3.6 do not enter: only the open orbit is normalized. Theorem 3.2.2 (multiplicity \(g\cdot a\)) does not enter either. For symmetric \(X\), \(W_X'\cap W_L=1\): the type-\(G\) vanishing \(T_\alpha S_\chi=0\) (\(\alpha\in\Delta(X)\)) of Theorem 3.7.1 is not needed, because those \(w\) already fail \(w(D)=D\).

Builds on

  • Proposition 1.5 · \(S_\chi\) over \(\mathbb F_q((t))\): convergence, generic uniqueness, rationality

Built on by

  • Computation 1.7 · \(M_\theta\): \(W_X'\cong S_3\), coboundary \(\beta\), formula (7.4), triangularity
  • Dead end 1.8 · \(B_{w_\gamma}\) for \(M_\theta\) over \(\mathbb F_q((t))\): two routes, missing lemmas
  • Theorem 1.9 · Theorem 8.0.2 over \(\mathbb F_q((t))\), \(p\) odd, symmetric \(X\): hypotheses
  • Audit 5.15 · Audit of A1-6: F_w S_χ = b_w S_{wχ}, the sum over W_X′, and the open-orbit route
  • Lemma S1.3 · Positivity of \(\hat S\) forces \(c\neq0\) and \(a_J/c>0\) over every local field
  • Theorem S1.5 · Theorem 8.0.2 over \(\mathbb F_q((t))\) for symmetric \(X\): the complete hypothesis list

Sources