§5 · Agent 5 · Audit: transfer, hypotheses · 2026-10-18T04:00:45Z

Audit 5.3 · Audit of A4-5: orbit count over \(\mathbb F_q((t))\), uniform \(D(\lambda)\)

proved audit: correct ★ key step

Second audit: correct. I re-derived Proposition 4.5 step by step ((E1)-(E4), the H-Cartan bound on h, K(D)-bi-invariance of the relation, the finite label set, and the injection between finite sets of equal size) and confirmed from S4b.json that S4-12/S4-13 assume only 'F p-adic, p odd', and from the Delorme-Secherre v1 text that the standing hypothesis is odd residue characteristic, so (H-Cartan) holds over every F'_e. The gap Audit 5.3 flags is real: Axiom 2.4.1Sakellaridis's Axiom 2.4.1, relabelled (A3b): the Cartan decomposition of X(F) into K-orbits. is phrased through A_X^+ (labels in the valuation monoid Lambda_X^+), which Proposition 4.5 never identifies in characteristic p. The repair via Proposition 3.7(3) and the dictionary of Theorem 3.8 is valid, and Proposition 4.6, Theorem 4.8 use only K\X(F) = Lambda_S^+. Correct.

The counting transfer of the Cartan decomposition is sound. Every condition is finitely determined with a bound independent of the field, and (H-Cartan) is available over every ramified \(F'_e\). The one gap: the claim 'Axiom 2.4.1Sakellaridis's Axiom 2.4.1, relabelled (A3b): the Cartan decomposition of X(F) into K-orbits. holds' additionally needs the characteristic-\(p\) valuation cone.

Audit 5.3.

Outcome: holds, with one repair of the conclusion's wording. (i) Steps 1–8 of Proposition 4.5 prove \(X(F)=\bigsqcup_{\kappa\in\Lambda_S^+}Kx_\kappa\) and \(G(F)=K\,T(F)\,H(F)\) for \(F=\mathbb F_q(\!(t)\!)\), \(p\) odd (\(p=3\) included). The inputs are (E1)–(E4) on both sides and (H-Cartan) over \(F'_e\) for every \(e\ge D(\lambda)\). (ii) Every condition in steps 4–6 is of type Lemma 4.2(e) with \(D\) independent of \(E\). The quantifier \(\exists h\in H(E)\) is replaced by \(\exists\eta\in B(\lambda),\ h_1,h_2\in H(\mathcal O)\), where \(B(\lambda)\) is finite and independent of \(E\). (iii) (H-Cartan) is needed over infinitely many \(F'_e\) (one \(e\ge D(\lambda)\) for each \(\lambda\)). S4-12/S4-13 give it for every \(p\)-adic field with \(p\) odd, never using \(e=1\). Proposition 3.4(d) with Audit 3.5 gives it directly for every local field of residue characteristic \(\ne2\). (iv) Gap. "Axiom 2.4.1Sakellaridis's Axiom 2.4.1, relabelled (A3b): the Cartan decomposition of X(F) into K-orbits. holds for \(\mathcal M_\theta\) over \(\mathbb F_q[\![t]\!]\)" is stronger than the orbit parametrization. \(A_X^+\) is defined through \(\Lambda_X^+\), the monoid of \(G\)-invariant valuations of \(X\) in characteristic \(p\) (checklist item 8). Proposition 4.5, which uses "no statement over \(\overline{\mathbb F_q(\!(t)\!)}\)", does not identify \(\Lambda_X^+\) with the image of \(\Lambda_S^+\). Repair: cite Proposition 3.7(3) (\(\Lambda_X^+=\Lambda_X\cap C^-\) in characteristic \(\ne2\)) and the dictionary of Theorem 3.8 step 1, or state only \(K\backslash X(F)\simeq\Lambda_S^+\).

  1. Inputs over both fields. (E1) Cartan decompositions of \(G(E)\) and of \(H(E)\). \(H_\Lambda\) is reductive with connected \(F_4\) fibres (Proposition 3.2) and contains the \(\Lambda\)-split maximal torus \(T_8\) (Audit 5.2 step 6), so it is split over the local ring \(\Lambda\), and Bruhat–Tits applies over every local field. (E2) \(X(\mathcal O_E)=Kx_0\): Hensel plus Lang for the smooth connected \(H\). (E3) \(X(E)=G(E)x_0\): \(G(E)\)-orbits are open (orbit maps are smooth, so they are open on \(E\)-points by the implicit function theorem, valid in every characteristic), and \(\mathring X(E)=B(E)x_0\) is dense because \(X\) is smooth and irreducible. Alternatively use \(H^1(F,F_4)=1\) (Lemma 3.3). (E4) \(\tau\) is defined over \(\Lambda\) and \(\theta(K)=K\).
  2. Step 3 re-derived. \(t^\lambda kx_0=a\,t^\lambda k'x_0\) with \(a\in K\) holds iff \(h:=(t^\lambda k)^{-1}at^\lambda k'\in H(E)\), i.e. iff \(t^\lambda khk'^{-1}t^{-\lambda}\in K\). This uses that the stabilizer of \(x_0\) in \(G(E)\) is \(H(E)\), as \(G\to X\) is an \(H\)-torsor.
  3. Step 4: bounding \(h\). \(h\in Kt^{-\lambda}Kt^\lambda K\subseteq\bigcup_{c\preceq\lambda+\lambda^*}Kt^cK\) by Lemma 4.2(c). If \(h=h_1t^\eta h_2\) with \(h_i\in H(\mathcal O)\subseteq K\), then \(h\in Kt^{\eta^+}K\), since a Tits lift in \(G(\mathbb Z)\) conjugates \(t^\eta\) to \(t^{\eta^+}\). So \(\eta^+\preceq\lambda+\lambda^*\). The map \(\eta\mapsto\eta^+\) is finite-to-one, so \(B(\lambda)\) is finite and does not depend on \(E\).
  4. Uniform \(D\). For each \(\eta\in B(\lambda)\), the matrix \(\rho_i(t^\lambda kh_1t^\eta h_2k'^{-1}t^{-\lambda})\) has denominators at most \(\varpi^{N_\eta}\), with \(N_\eta\) fixed by \(\lambda,\eta\). The condition "\(\operatorname{val}=0\) for all \(i\)" is determined modulo \(\varpi^{N_\eta+1}\). \(D_1(\lambda)=\max_\eta(N_\eta+1)\). The relation \(\sim\) on \(K\) depends only on reductions modulo \(\varpi^{D_1}\), so it is the pull-back of an equivalence relation on \(G(\mathcal O/\varpi^{D_1})\) whose defining polynomials have coefficients in \(\Lambda\). Under \(D_1\)-closeness the two relations correspond (the sets \(H(\mathcal O/\varpi^{D_1})\) correspond by Lemma 4.2(b)). The same holds for \(M_\lambda\) in step 6, with \(\kappa\) in the finite set \(\{\kappa\preceq\lambda+(-\theta\lambda)^+\}\).
  5. Counting (step 7). Over \(E'\), \(\kappa\mapsto Kx_\kappa\) is a bijection \(M_\lambda(E')\to K\backslash Z_\lambda(E')\): injective by step 1, onto by (H-Cartan)\(_{E'}\) restricted to \(Z_\lambda(E')\). Over \(E\) it is injective with \(|M_\lambda(E)|=N_\lambda(E)\), hence onto. \(N_\lambda(E)<\infty\) because \(K\)-orbits are open and \(Z_\lambda=\bigcup_{gK\subseteq Kt^\lambda K}gKx_0\) is compact.
  6. (H-Cartan) over ramified fields. S4-12 applies Delorme–Sécherre's Theorem 4.8, whose standing hypothesis is a locally compact non-archimedean field of residue characteristic \(\ne2\) (read in the arXiv text). \(|J|=1\) follows from Kneser's \(H^1(F',\mathrm{Spin}_8)=1\) (every finite extension of \(\mathbb Q_p\)) together with Helminck–Wang and Vust–Richardson, all characteristic-\(0\) results applied in characteristic \(0\) (checklist item 1 respected). S4-13 realizes \(W_X\) by \(\dot n_i\in H(\mathbb Z)\subseteq K'\). No step uses \(e(F'/\mathbb Q_p)=1\). Proposition 3.4 proves \(|J|=1\) for every local field with \(p\ne2\) without these citations.
  7. The gap (iv). Axiom 2.4.1Sakellaridis's Axiom 2.4.1, relabelled (A3b): the Cartan decomposition of X(F) into K-orbits. (p. 10) speaks of \(A_X^+\), the points whose label lies in \(\Lambda_X^+\). That \(\Lambda_X^+\) corresponds to \(\Lambda_S^+\) under \(\Lambda_S\cong\Lambda_X\) is a statement about invariant valuations of \(X_{\bar F}\), i.e. Knop's §5 in characteristic \(0\). In characteristic \(p\) it is Proposition 3.7(3) (re-checked in Objection 5.11). Proposition 4.5 does not supply it, so its sentence "so Axiom 2.4.1Sakellaridis's Axiom 2.4.1, relabelled (A3b): the Cartan decomposition of X(F) into K-orbits. holds" needs that citation. The parametrization \(K\backslash X(F)\simeq\Lambda_S^+\), which is all that Proposition 4.6 and Theorem 4.8 use, is unaffected.

Builds on

  • Proposition 4.5 · Proposition: \(K\backslash X(\mathbb F_q((t)))\simeq\Lambda_S^+\) by counting orbits

Built on by

  • Summary 5.13 · Verdict of audit A5: transfer route A4-2…A4-9 and char-p hypotheses A3-1…A3-10

Sources