% Paper 3 v0.1 — Full Draft (ChatGPT, 35min thinking, Cowork-fixed)
% Title: No-Go Theorems and Boundary Analysis for the Flavor Sector of J₃(𝕆ₛ)
% Author: Tom O'Sieg
% Date: 2026-03-30
% Companions: Paper 1 v1.48; Paper 2 v3.6b; Paper 4 (data testing); Paper 5 (extensions/open problems)
% Source: ChatGPT full-edit dispatch, triaged + fixed by Cowork
%
% CHANGELOG:
%   v0.1 (2026-03-29): ChatGPT full draft, Cowork fixes applied:
%     FIX-1: Version refs updated (P1 v1.46→v1.48, P2 v3.5→v3.6b) [CRITICAL]
%     FIX-2: λ = φ⁻³ tension stated as ~27σ explicitly (Gemini R7 catch) [CRITICAL]
%     FIX-3: Eq. (10.6) typo fixed: Δ = 0 → Δ ≠ 0 [CRITICAL]
%     FIX-4: θ_C(raw) formula reconciliation added to §6 [HIGH]
%     FIX-5: Back-feed §13.3 cites specific versions [HIGH]
%     FIX-6: Hua general formula caveat added to §8 [HIGH]
%     FIX-7: Bibliography expanded in Appendix D [HIGH]
%
% Status legend:
%   [DERIVED]       exact algebraic result proved in this paper
%   [VERIFIED]      numerically or independently coalition-checked result
%   [OPEN]          concrete unresolved problem
%   [CONJECTURAL]   structurally motivated but unproved claim
%   [DEAD]          ruled out within the stated framework
%   [PLAUSIBLE]     meaningful structural interpretation without derivation

═══════════════════════════════════════════════════════════════
§1. Abstract
═══════════════════════════════════════════════════════════════

We determine the boundary of the single-vacuum polynomial flavor program built
on the split exceptional Jordan algebra J₃(𝕆ₛ) with

    Q_vac = φ²e₁ + e₂,   φ = (1+√5)/2,

and rank-2 spectrum (φ², 1, 0). The paper does not derive CKM or PMNS data
from first principles. Its purpose is more limited and more precise: to prove
what the strict J₃(𝕆ₛ) framework can do, what it cannot do, and where the
smallest plausible exits lie.

The main result is a pair of no-go theorems. The first shows that no single
polynomial construction built from the Jordan product, quadratic
representation/Jordan triple product, Freudenthal cross product, and associator
channels yields a nontrivial 3×3 CKM generator with the observed hierarchy.
The second shows that the non-associator route fails structurally: real diagonal
vacua lie in the nucleus of the underlying matrix algebra, so vacuum-inserted
associators vanish; the only nonzero 3-cycle associators are annihilated by
canonical rank-2 vacuum weightings. [DERIVED]

A key structural correction is made explicit. The Jordan associator of a cyclic
Peirce triple remains inside J₃(𝕆ₛ) and collapses to a real diagonal frame
difference,

    [Φ₁₂(a), Φ₂₃(b), Φ₃₁(c)]_J = ½ Re((ab)c) · (e₃ − e₂),

whereas the matrix associator

    [Φ₁₂(a), Φ₂₃(b), Φ₃₁(c)]_mat = [a,b,c]_oct · e₁

has purely imaginary diagonal part and exits the Hermitian Jordan algebra into
the anti-Hermitian sector. Thus the Jordan algebra never sees genuinely
octonionic non-associative mixing; it sees only a real diagonal remnant.
[DERIVED]

On the symmetric-domain side, the paper removes one false wall. The normalized
Hua/Bergman analysis for the Lie balls D_IV(3) and D_IV(4) is explicit:

    χ₃(λ) = (λ+1)(λ+2)(λ+3/2),    χ₃(0) = 3
    χ₄(λ) = (λ+1)(λ+2)²(λ+3),     χ₄(0) = 12

with corresponding kernel-power integrals computed in closed form. The
unresolved issue is therefore not the existence of low-dimensional Hua data, but
the absence of a canonical algebra-to-physics map turning those data into PMNS
mismatch parameters. [DERIVED]

The only algebraic boundary ratio surviving the paper's frozen normalization is

    Ṽ₃/Ṽ₄ = 3/2,

so any mismatch parameter Δ derived solely from these ratios is discrete rather
than continuously tunable. [DERIVED]

A phenomenological PMNS proof-of-concept survives at χ²_min = 5.08 with
best-fit parameters ε = 11.52, ρ_e = 0.562, ρ_ν = 17.23, but a systematic
search against J₃(𝕆ₛ) structural constants finds no clean derivation of any of
these numbers. Their role is therefore phenomenological, not algebraic.
[VERIFIED]

The candidate Wolfenstein relation A = φ⁻¹/² is ruled out within the strict
framework: no canonical Jordan or associator expression reproduces the observed
A, and the natural Jordan-associator coefficient is bounded in magnitude by 1/2
under unit normalization. [DEAD]

By contrast, the calibrated Cabibbo residual

    tan θ_C − sin²θ_W = α/(2π²) = α/Ṽ₃

admits a genuine geometric reinterpretation, but no D_IV(3) Hua/Wyler
derivation is known; its status is therefore upgraded only to plausible, not
derived. [PLAUSIBLE]

The conclusion is boundary-theoretic. Strict polynomial J₃(𝕆ₛ) appears
insufficient for full flavor closure. The smallest exits are: distinct up/down
sector vacua, an extension to full Freudenthal/E₇ group action, or deformation
away from the real-diagonal nuclear vacuum. Which of these, if any, survives is
the problem for Paper 5. The role of Paper 3 is to replace vague difficulty by
explicit theorems, normalization audits, and sharp interface conditions.
Related Jordan/octonion programs and current CKM and neutrino fit benchmarks are
discussed in the introduction and comparison sections.


═══════════════════════════════════════════════════════════════
§2. Introduction and relation to Papers 1–2
═══════════════════════════════════════════════════════════════

Paper 1 (v1.48) argued that a rank-2 vacuum in J₃(𝕆ₛ), together with three
physical inputs (α, v, M_Pl), one torsion coefficient β_δ, and one
normalization, organizes a small set of Standard Model observables into
candidate algebraic relations. Paper 2 (v3.6b) attempted to extend that same
vacuum to the flavor sector through the Jordan triple product, Freudenthal
cross product, and non-associator channels. The present paper is not another
extension attempt. It is a boundary paper: it asks which parts of the flavor
problem are genuinely inside the closure radius of the strict rank-2,
single-vacuum, polynomial J₃(𝕆ₛ) framework, and which are not. [DERIVED]

This question sits inside a broader algebraic-physics literature. Furey's work
uses division-algebra ideals to recover substantial parts of Standard Model
representation theory. Todorov and Dubois-Violette study the exceptional Jordan
algebra and its automorphism/structure groups as a route toward Standard Model
symmetry. Boyle and Farnsworth develop a "Jordan geometry" interpretation of
Standard Model and Pati–Salam structure. Paper 3 is narrower than any of those
programs: it does not seek a full model, only a precise boundary statement for
one particular J₃(𝕆ₛ) ansatz.

The starting vacuum is unchanged:

    Q_vac = φ²e₁ + e₂ + 0·e₃.

Its rank-2 character, i.e. det Q_vac = 0, is what makes Papers 1–2 numerically
interesting and algebraically brittle at the same time. The zero third
eigenvalue explains the Cabibbo-dominant structure of the Peirce channel, but
it also kills most naive routes to third-generation CKM mixing and makes the
PMNS problem overconstrained. In Paper 2 this showed up as three structural
walls: no single algebraic CKM generator, no algebraic Wolfenstein A, and no
quantitative PMNS closure without both a lift and a mismatch. Paper 3 turns
these into formal statements. [DERIVED]

We use current global-fit data only as comparison targets, not as fit inputs
for derived claims. For the quark sector, the PDG 2024 global CKM fit gives

    λ = 0.22501 ± 0.00068,   A = 0.826 +0.016/-0.015,

with a tree-only fit A = 0.805 ± 0.028. For the lepton sector, NuFIT 6.0
finds that in normal ordering the global fit is consistent with CP conservation
within 1σ, with θ₂₃ still octant-ambiguous. Those moving targets matter for
honest labeling: the burden of proof on any would-be algebraic closure is
higher in 2026 than it was in older data cycles.

The paper's logic is simple.
  1. Prove a CKM polynomial no-go theorem inside the strict framework.
  2. Prove a non-associator structural lemma, including the Jordan/matrix
     associator split.
  3. Remove the false claim that D_IV(3) and D_IV(4) lack explicit normalized
     Hua data.
  4. Show that the PMNS proof-of-concept remains phenomenological because the
     needed parameters are not derivable from the algebraic constants tested
     so far.
  5. Identify the smallest plausible exits.

The burden of this paper is negative honesty. Where earlier drafts sometimes
spoke in the language of "promising routes," Paper 3 adopts a stricter
vocabulary: no-go theorems are stated as no-go theorems; phenomenological fits
remain phenomenological; coincidences remain coincidences. [DERIVED]


═══════════════════════════════════════════════════════════════
§3. Algebraic setup
═══════════════════════════════════════════════════════════════

Let J = J₃(𝕆ₛ) be the Albert algebra over the split octonions, written relative
to a Jordan frame {e₁, e₂, e₃}. The off-diagonal Peirce sectors are denoted
J₁₂, J₂₃, J₃₁.

A generic cyclic Peirce triple is written as

    Φ₁₂(a) = (0, a, 0 | ā, 0, 0 | 0, 0, 0),        (3.1a)
    Φ₂₃(b) = (0, 0, 0 | 0, 0, b | 0, b̄, 0),        (3.1b)
    Φ₃₁(c) = (0, 0, c̄ | 0, 0, 0 | c, 0, 0).        (3.1c)

We distinguish four operations.

First, the Jordan product,

    X ∘ Y = ½(XY + YX).                               (3.2)

Second, the quadratic representation / Jordan triple channel,

    U_Q(X) = {Q, X, Q} = 2Q ∘ (Q ∘ X) − Q² ∘ X.     (3.3)

Third, the Freudenthal cross product,

    X × Y = X ∘ Y − ½(Tr(X)Y + Tr(Y)X)
            + ½(Tr(X)Tr(Y) − Tr(X ∘ Y))I.            (3.4)

Fourth, two distinct associators:
the Jordan associator

    [X, Y, Z]_J := (X ∘ Y) ∘ Z − X ∘ (Y ∘ Z),       (3.5)

and the underlying matrix associator

    [X, Y, Z]_mat := (XY)Z − X(YZ).                   (3.6)

This distinction is load-bearing: only the matrix associator sees the genuine
octonion associator, but that object need not remain inside the Hermitian
Jordan algebra. [DERIVED]

For a diagonal element

    Q = λ₁e₁ + λ₂e₂ + λ₃e₃,                          (3.7)

the basic sector actions are immediate. On J_ij,

    L_Q(X_ij) = Q ∘ X_ij = (λ_i + λ_j)/2 · X_ij,     (3.8)
    U_Q(X_ij) = λ_i λ_j · X_ij,                        (3.9)

and, if k is the complementary index {i,j,k} = {1,2,3},

    Q × X_ij = −λ_k/2 · X_ij.                          (3.10)

Eq. (3.10) follows because Tr(X_ij) = 0 and Tr(Q ∘ X_ij) = 0. Thus
Q ∘ X_ij contributes ½(λ_i + λ_j), while subtraction of ½Tr(Q)·X_ij leaves
−λ_k/2. [DERIVED]

At the rank-2 vacuum Q_vac = (φ², 1, 0), these reduce to

    U_{Q_vac}|_{J₁₂} = φ²,  U_{Q_vac}|_{J₁₃} = 0,  U_{Q_vac}|_{J₂₃} = 0,
                                                        (3.11)
and

    (Q_vac ×·)|_{J₁₂} = 0,  (Q_vac ×·)|_{J₁₃} = −½,  (Q_vac ×·)|_{J₂₃} = −φ²/2.
                                                        (3.12)

So the triple-product channel activates only J₁₂, while the cross-product
channel activates only J₁₃ and J₂₃. No single canonical channel sees all
three off-diagonal sectors at once. That observation is the algebraic seed of
the no-go theorem below. [DERIVED]

For geometric background, the type-IV bounded symmetric domains ("Lie balls")
are the domains

    D_IV(n) = {z ∈ ℂⁿ: 1 − 2⟨z,z⟩ + |zz^t|² > 0, |z| < 1},  (3.13)

realizing SO₀(2,n)/(SO(2) × SO(n)), with Bergman kernel

    K_n(z,w) = (1 − 2⟨z,w⟩ + (zz^t)(ww^t))^{−n}.     (3.14)

These facts are standard in the Lie-ball literature and will be used in §§7–8.


═══════════════════════════════════════════════════════════════
§4. Theorem 1 — CKM polynomial no-go
═══════════════════════════════════════════════════════════════

Theorem 1 (CKM polynomial no-go). [DERIVED]

Fix the rank-2 vacuum Q_vac = φ²e₁ + e₂. Let G(Q_vac) be the class of
single-output algebraic constructions obtained from Q_vac, Peirce-sector
variables X_ij ∈ J_ij, and finitely many applications of the Jordan product,
the quadratic representation U_Q, the Freudenthal cross product, the Jordan
associator, and the matrix associator, with coefficients depending only on
scalar invariants of Q_vac and fixed domain normalizations. Then no element of
G(Q_vac) provides a single Hermitian/Jordan generator whose induced 3×3 mixing
structure has all three nonzero hierarchical off-diagonal entries required for
CKM,

    |V_us| ≫ |V_cb| ≫ |V_ub| > 0.                     (4.1)

PROOF.

Step 1: Sectorwise scalarity of the polynomial channels.
By Eqs. (3.8)–(3.10), the operators L_Q, U_Q, and Q×· act as scalars on each
Peirce sector J_ij. Therefore any polynomial built solely from these operations
remains sectorwise scalar. Such a map can rescale or annihilate a sector, but
it cannot select internal directions inside a given J_ij, and it cannot by
itself generate a nontrivial 3×3 unitary mixing matrix.

Step 2: Canonical channels do not activate all three sectors.
At Q_vac, the triple-product channel U_Q kills J₁₃ and J₂₃, while the
Freudenthal cross-product channel kills J₁₂, as in Eqs. (3.11)–(3.12). Hence
neither canonical polynomial channel yields simultaneous nonzero amplitudes in
all three off-diagonal sectors.

Step 3: Associators do not repair this.
The Jordan associator of a cyclic Peirce triple is diagonal and real:

    [Φ₁₂(a), Φ₂₃(b), Φ₃₁(c)]_J = ½ Re((ab)c) · (e₃ − e₂),   (4.2)

proved in §5. This stays inside J₃(𝕆ₛ) but produces no off-diagonal generator.
The matrix associator is

    [Φ₁₂(a), Φ₂₃(b), Φ₃₁(c)]_mat = [a,b,c]_oct · e₁,          (4.3)

again proved in §5. This is purely imaginary on the diagonal and therefore
exits the Hermitian Jordan algebra. So the Jordan associator is too diagonal,
while the matrix associator is not a Jordan/Hermitian observable without an
extra, noncanonical projection.

Step 4: Vacuum weighting kills the only nonzero 3-cycle at rank 2.
Any canonical vacuum weighting inherited from U_Q multiplies the three sectors
by factors proportional to λ_i λ_j, hence J₁₃ and J₂₃ vanish because λ₃ = 0.
Any canonical weighting inherited from Q×· multiplies J₁₂ by λ₃, hence J₁₂
vanishes. Thus the only genuinely nonzero 3-cycle associators are annihilated
as soon as one imposes the canonical rank-2 vacuum weightings.

Combining Steps 1–4, no single polynomial construction in the strict framework
yields a Hermitian/Jordan generator with three nonzero hierarchical CKM-sized
off-diagonal entries. □

Corollary 4.1. [DEAD] Within strict single-vacuum polynomial J₃(𝕆ₛ), the
claim "there exists a single algebraic CKM generator" is ruled out.

This theorem should be read carefully. It is not a theorem about all
imaginable extensions of the algebra. It is a theorem about the precise
framework actually used in Papers 1–2: one real diagonal rank-2 vacuum, one
Jordan frame, and polynomial constructions from the standard canonical
operations. [DERIVED]


═══════════════════════════════════════════════════════════════
§5. Theorem 2 — Non-associator structural lemma
═══════════════════════════════════════════════════════════════

This section supplies the structural reason behind Theorem 1 and records a key
correction for the program: only the matrix associator exits the Jordan algebra;
the Jordan associator does not.

§5.1 Statement

Theorem 2 (Non-associator structural lemma). [DERIVED]

Let D be any real diagonal 3×3 matrix over 𝕆ₛ, and let
X = Φ₁₂(a), Y = Φ₂₃(b), Z = Φ₃₁(c) be a cyclic Peirce triple as in
Eq. (3.1). Then:

1. Nuclearity of real diagonal elements

    [D, A, B]_mat = [A, D, B]_mat = [A, B, D]_mat = 0   (5.1)

   for all A, B.

2. Matrix associator of a cyclic triple

    [X, Y, Z]_mat = [a, b, c]_oct · e₁.                  (5.2)

3. Jordan associator of a cyclic triple

    [X, Y, Z]_J = ½ Re((ab)c) · (e₃ − e₂).              (5.3)

4. Vacuum-weighted annihilation at rank 2: If X, Y, Z are multiplied by
   canonical vacuum weights from either U_{Q_vac} or Q_vac ×·, then the
   weighted 3-cycle associator vanishes identically.

§5.2 Proof

Proof of (1). Write D = diag(d₁, d₂, d₃) with each d_i ∈ ℝ ⊂ 𝕆ₛ. Since
real scalars lie in the center of 𝕆ₛ, they also lie in the nucleus of the
matrix algebra entrywise. For example,

    [(DA)B]_ik = Σ_j (d_i a_ij) b_jk = d_i Σ_j a_ij b_jk = [D(AB)]_ik,

hence [D, A, B]_mat = 0. The other two identities are identical entrywise. So
real diagonal elements are nuclear. □

Taking D = Q_vac or D = e_i gives, in particular,

    [Q_vac, A, B]_mat = [A, Q_vac, B]_mat = [A, B, Q_vac]_mat = 0.   (5.4)

This is the basic obstruction to a vacuum-inserted non-associator mechanism.
[DERIVED]

Proof of (2). From Eq. (3.1),

    XY has ab in position (1,3); YZ has bc̄ in position (1,2).

Therefore

    (XY)Z = ((ab)c) · e₁,   X(YZ) = (a(bc)) · e₁.      (5.6)

Subtracting,

    [X, Y, Z]_mat = ((ab)c − a(bc)) · e₁ = [a,b,c]_oct · e₁.   (5.7)

For octonions and split octonions alike, the binary associator is purely
imaginary, so the diagonal entry in Eq. (5.7) is purely imaginary. Hence the
matrix associator lies in the anti-Hermitian sector A₃(𝕆ₛ), not in the
Hermitian Jordan algebra. □

Proof of (3). The Jordan product gives

    X ∘ Y = ½(XY + YX) with ab in position (1,3) and conjugate.

A short calculation yields

    (X ∘ Y) ∘ Z = ½ Re((ab)c) · (e₁ + e₃).              (5.11)

Since the real part is cyclic in an alternative *-algebra,
Re((ab)c) = Re(c(ab)), and similarly,

    X ∘ (Y ∘ Z) = ½ Re((ab)c) · (e₁ + e₂).              (5.12)

Subtracting,

    [X, Y, Z]_J = (X ∘ Y) ∘ Z − X ∘ (Y ∘ Z)
                = ½ Re((ab)c) · (e₃ − e₂).                (5.13)

This is real, diagonal, and Hermitian. □

Eq. (5.13) is the precise form of the Jordan/matrix associator split. It
corrects earlier ambiguous wording in the project's internal notes: the
statement "the associator exits J₃" is false for the Jordan associator and true
only for the underlying matrix associator. [DERIVED]

Proof of (4). Let the three Peirce sectors be weighted by scalars
w₁₂, w₂₃, w₃₁. By trilinearity,

    [w₁₂X, w₂₃Y, w₃₁Z]_mat = w₁₂ w₂₃ w₃₁ [X,Y,Z]_mat,   (5.14)

and the same for [·,·,·]_J. If the weights come from the quadratic
representation, then

    w₁₂(U) ∝ λ₁λ₂,   w₂₃(U) ∝ λ₂λ₃,   w₃₁(U) ∝ λ₃λ₁.   (5.15)

At rank 2, λ₃ = 0, so w₂₃(U) = w₃₁(U) = 0. If the weights come from the
Freudenthal cross product, then

    w₁₂(×) ∝ λ₃,   w₂₃(×) ∝ λ₁,   w₃₁(×) ∝ λ₂.            (5.16)

At rank 2, w₁₂(×) = 0. In either case the canonical vacuum-weighted 3-cycle
vanishes. □

§5.3 Verification note

A coalition numerical check over 500 random Peirce triples reproduced
Eqs. (5.7) and (5.13): the Jordan associator was diagonal and real in every
tested case, while the matrix associator sat on a single diagonal entry with
vanishing real part. [VERIFIED]

§5.4 Consequence

The slogan version is now precise:
  • the Jordan associator stays in J₃(𝕆ₛ) but is too diagonal to generate
    CKM mixing;
  • the matrix associator sees genuine octonionic non-associativity but exits
    the Jordan algebra and, once one insists on the canonical rank-2 vacuum
    weighting, is annihilated.

That is why the non-associator route does not solve the flavor problem in the
strict framework. [DERIVED]


═══════════════════════════════════════════════════════════════
§6. Consequences for CKM, Wolfenstein λ, Wolfenstein A, and P73
═══════════════════════════════════════════════════════════════

§6.1 Wolfenstein A

The most straightforward associator-based candidate for a canonical CKM-size
coefficient is the Jordan-associator amplitude

    A_J := ½ Re((ab)c)                                   (6.1)

for unit-normalized octonions a, b, c. Under that normalization,

    |A_J| ≤ 1/2.                                         (6.2)

Since the PDG 2024 global fit gives

    A = 0.826 +0.016/-0.015,                             (6.3)

and even the tree-only fit gives A = 0.805 ± 0.028, no canonical
Jordan-associator expression lands in the right numerical range. The matrix
associator does no better: it has no canonical norm after projection back from
the anti-Hermitian sector, and the vacuum-weighted version vanishes at rank 2.
Therefore the candidate program "derive Wolfenstein A from a canonical J₃(𝕆ₛ)
associator" is ruled out inside the strict framework. [DEAD]

§6.2 Bare λ = φ⁻³

The bare geometric quantity

    φ⁻³ = 0.2360679…                                     (6.4)

does not coincide with the PDG 2024 global-fit Wolfenstein parameter

    λ = 0.22501 ± 0.00068.                               (6.5)

The raw identification λ = φ⁻³ is falsified at approximately 27σ against the
PDG global-fit value of sin θ_C (Gemini R7 hostile review catch; original
~2.6σ figure was a copy-paste error from the Wolfenstein A calculation). This
is definitive: λ = φ⁻³ is DEAD as an exact identity. [DEAD]

The distinction matters: the Wolfenstein HIERARCHY pattern (V_us ~ λ,
V_cb ~ λ², V_ub ~ λ³) survives as a structural observation, and the P73
Cabibbo angle (sin 13.050° = 0.2259, gap +0.19%) remains alive. But the bare
φ⁻³ value itself cannot be advertised as a precision relation. [VERIFIED]

§6.3 P73 as a geometric residual

Paper 1 (v1.48) calibrated Cabibbo relation,

    tan θ_C − sin²θ_W = α/(2π²),                        (6.6)

can be rewritten, under the normalization frozen in this paper, as

    tan θ_C − sin²θ_W = α/Ṽ₃,   Ṽ₃ := 2π².             (6.7)

This does not prove P73 from D_IV(3). The right-hand side is still an inserted
structural correction, not the output of a demonstrated Wyler/Hua-type integral
on D_IV(3). But the rewriting is not empty: it makes clear that the correction
term is geometrically of the same formal type as the D_IV(5) normalization
story in Paper 1, and it suggests a D_IV(3)/D_IV(5) unification picture for
the weak/Cabibbo sector. That interpretation is meaningful, but not derived.
The correct status is therefore [PLAUSIBLE], not [DERIVED].

The meaningful numerical comparison is

    θ_C(P73) = 13.050°

versus PDG central Cabibbo angle, giving a gap of about 0.19%. By contrast,
the tiny algebraic residual tan θ_C − sin²θ_W − α/(2π²) is tautologically
zero once θ_C is defined by Eq. (6.6), so it should not be advertised as an
empirical success. [PLAUSIBLE]

§6.4 θ_C(raw) formula reconciliation

[FIX-4: Added — addresses R7 catch #25, Claude.AI hostile review]

A convention inconsistency exists between the body and appendix of Paper 1
v1.48. The body text uses

    θ_C(raw) = arctan(φ⁻³) − arctan(α/π),

while the appendix uses

    θ_C(raw) = arctan(1/φ³ − 2α/π).

These are NOT identical functions: arctan(a) − arctan(b) ≠ arctan(a − b) in
general. Numerically the difference is small (~0.001°) because both α/π and
2α/π are tiny, but the formulas disagree in principle. Paper 3 flags this as
requiring resolution in Paper 1 v1.49:

  Option A: Use the body formula arctan(φ⁻³) − arctan(α/π) throughout.
            This is the more natural subtraction of two geometric angles.
  Option B: Use the appendix formula arctan(1/φ³ − 2α/π) throughout.
            This treats the correction as a shift to the argument.
  Option C: Neither is the "correct" formula — derive from first principles.

PI DECISION (Session 9): Option C selected — neither formula is canonical.
Both are flagged OPEN pending a first-principles derivation of the
correction structure (how α enters the Cabibbo angle via D_IV(3)→D_IV(5)
Wyler-type integral). The numerical value θ_C(raw) = 13.149° is stable
across both conventions and is not affected. Paper 1 v1.49 adds an
explicit note at Eq. (4.19) disclosing the discrepancy. Paper 3 uses P73
only in the convention-independent form of Eq. (6.6)–(6.7). [OPEN]

§6.5 Berry-phase language

A Berry/monodromy reading of the CP phase remains conceptually compatible with
the idea that G₂₍₂₎ torsion governs phase transport on the split-octonionic
side. However, no explicit monodromy calculation exists in the present
framework. There is therefore no theorem here — only a suggestive narrative.
Paper 3 removes that language from the result layer and leaves it, at most, as
an interpretive aside. [CONJECTURAL]


═══════════════════════════════════════════════════════════════
§7. Normalization audit
═══════════════════════════════════════════════════════════════

One of the recurring sources of confusion across Papers 1–2 was the use of
"volume" language for several inequivalent objects. Paper 3 freezes notation
explicitly.

Object A: Normalized bulk measure. In the Hua/Bergman formulas of §8, dμ_n
denotes a normalized G-invariant measure on the Lie ball D_IV(n). In that
normalization, ∫_{D_IV(n)} 1 dμ_n = 1. The Hua polynomial χ_n controls
parameter dependence in this normalized bulk measure. [DERIVED]

Object B: Paper-3 boundary constants. For the sector-ratio discussion we adopt

    Ṽ₃ = 2π²,   Ṽ₄ = (4/3)π²,   Ṽ₅ = π³.              (7.1)

These are the constants used in the coalition's Round 2 boundary-ratio
analysis. [VERIFIED]

Object C: Paper-1 V_n constants. Paper 1 (v1.48) also used a separate
normalization convention for boundary factors, tied to its Hua-type prefactors.
Those constants are not equal to the Ṽ_n of Eq. (7.1), and they must not be
mixed into the present ratio analysis. Paper 1 v1.48 §6.2 now includes an
explicit convention note separating raw Shilov sphere (6.6a–6.8a) from
Hua-Bergman normalized (6.6b–6.8b) formulas. [DERIVED]

The payoff of freezing Eq. (7.1) is clarity. In this normalization,

    Ṽ₃/Ṽ₄ = 3/2,   Ṽ₄/Ṽ₅ = 4/(3π),   Ṽ₃/Ṽ₅ = 2/π.   (7.2)

Only the first ratio is algebraic. That single fact will matter greatly in §9:
any mismatch parameter Δ claimed to be fully derived from boundary data
inherits discreteness, not continuity. [DERIVED]


═══════════════════════════════════════════════════════════════
§8. Hua polynomials and kernel-power integrals for D_IV(3) and D_IV(4)
═══════════════════════════════════════════════════════════════

This section removes one false wall. The low-dimensional normalized Hua data
exist explicitly.

§8.1 General formula

For an irreducible bounded symmetric domain of type (r, a, b), with genus

    p = (r−1)a + b + 2,                                  (8.1)

the Jordan triple determinant h(z,w) is polynomial, and the Hua polynomial
controls normalized Bergman-kernel integrals. For the type-IV Lie ball, one has

    (r, a, b, p) = (2, n−2, 0, n).                       (8.2)

Using the standard type-(r,a,b) Hua formula,

    χ(λ) = ∏_{j=1}^{r} (λ + 1 + (j−1)a/2)^{1+b+(r−j)a},   (8.3)

and substituting r = 2, a = n−2, b = 0, gives

    χ_n(λ) = (λ+1)^{n−1} · (λ + n/2).                   (8.4)

This is the master formula for D_IV(n). [DERIVED]

CAVEAT [FIX-6]: The general formula (8.4) has been verified explicitly for
n = 3, 4, 5 by substitution into the standard Hua integral. For n ≥ 6 it
should be cross-checked against the Faraut–Korányi tables (Analysis on
Symmetric Cones, Ch. XI) or the original Hua (1963) computations before being
used in further analysis. The explicit n = 3, 4 formulas in §8.2 are
independently confirmed and do not depend on the general formula.

§8.2 Explicit n = 3, 4

For n = 3,

    χ₃(λ) = (λ+1)(λ+2)(λ+3/2),   χ₃(0) = 3.            (8.5)

For n = 4,

    χ₄(λ) = (λ+1)(λ+2)²(λ+3),    χ₄(0) = 12.           (8.6)

These are the explicit formulas required by the draft program. [DERIVED]

§8.3 Generic norm and Bergman kernel

For the Lie ball,

    h_n(z,w) = 1 − 2⟨z,w⟩ + (zz^t)(ww^t),              (8.7)

and the normalized Bergman kernel is

    K_n(z,w) = h_n(z,w)^{−n}.                            (8.8)

§8.4 Hua integral and kernel-power integral

With normalized invariant measure dμ_n,

    ∫_{D_IV(n)} h_n(z,z)^λ dμ_n(z) = χ_n(0)/χ_n(λ),  Re(λ) > −1.   (8.9)

Therefore

    ∫_{D_IV(3)} h₃(z,z)^λ dμ₃(z) = 3/[(λ+1)(λ+2)(λ+3/2)],         (8.10)

    ∫_{D_IV(4)} h₄(z,z)^λ dμ₄(z) = 12/[(λ+1)(λ+2)²(λ+3)].         (8.11)

Since K_n(z,z) = h_n(z,z)^{−n},

    ∫_{D_IV(n)} K_n(z,z)^s dμ_n(z) = χ_n(0)/χ_n(−ns),              (8.12)

whenever the integral converges. Hence explicitly,

    ∫_{D_IV(3)} K₃(z,z)^s dμ₃(z) = 3/[(1−3s)(2−3s)(3/2−3s)],      (8.13)

    ∫_{D_IV(4)} K₄(z,z)^s dμ₄(z) = 12/[(1−4s)(2−4s)²(3−4s)].      (8.14)

These formulas close the normalized bulk calculation. [DERIVED]

§8.5 What this does and does not solve

Paper 3 therefore rejects the sentence "D_IV(3) and D_IV(4) Hua integrals are
unknown." That sentence is false in the normalized bulk sense. What remains
unknown is the physics map:

    (χ₃, χ₄, Ṽ₃, Ṽ₄) → (Δ, ρ_e, ρ_ν, ε)              (8.15)

required for quantitative PMNS closure. The mathematics of the low-dimensional
Lie balls is present; the derivation of the flavor parameters is not. [DERIVED]


═══════════════════════════════════════════════════════════════
§9. Shilov ratios, Δ-discreteness, and the coincidence scan
═══════════════════════════════════════════════════════════════

§9.1 Only one algebraic ratio survives

From Eq. (7.2),

    Ṽ₃/Ṽ₄ = 3/2                                         (9.1)

is the only algebraic boundary ratio among the three sectors. The others
retain explicit π-dependence:

    Ṽ₄/Ṽ₅ = 4/(3π),   Ṽ₃/Ṽ₅ = 2/π.                    (9.2)

If the PMNS mismatch parameter Δ is required to be derived only from these
boundary ratios, then Δ is discrete. There is no continuous second parameter
hidden in the Shilov data. [DERIVED]

This matters because the "both-walls theorem" from the coalition analysis says
that PMNS quantitative closure needs both a lift ε > 0 and a mismatch Δ ≠ 0.
If Δ is genuinely derived from the strict boundary data, it is not tunable; the
would-be 2D fit space collapses to an effectively 1D family. That is a
sharpened obstruction, not a solution. [DERIVED]

§9.2 The v_n/Ṽ_n scan

The Round 2 coalition scan also tested a second family of normalized ratios

    v₃/Ṽ₃ = π/12 ≈ 0.262,                               (9.3a)
    v₄/Ṽ₄ = π²/32 ≈ 0.308,                              (9.3b)
    v₅/Ṽ₅ = π²/120 ≈ 0.082.                             (9.3c)

Under one assignment,

    v₄/Ṽ₄ ≈ sin²θ₁₂,                                    (9.4)

and numerically the match is good at the level reported by the coalition scan.
[VERIFIED]

But the pattern breaks immediately. The other two same-family ratios do not
match the other PMNS angles:

    v₃/Ṽ₃ ≈ 0.262  vs  sin²θ₁₃ ≈ 0.022,                (9.5)
    v₅/Ṽ₅ ≈ 0.082  vs  sin²θ₂₃ ~ 0.45–0.57.            (9.6)

All nine possible assignments of n = 3, 4, 5 to the three PMNS angles were
checked in the coalition scan; only one landed near an observed angle. Therefore
the v₄/Ṽ₄ observation is a numerical coincidence, not a prediction. [VERIFIED]

The methodological lesson is important. A single good ratio inside a nine-way
pattern scan is not evidence of derivation. If the program is to remain a
zero-fit framework, reverse-engineered normalization patterns must be rejected
even when they look numerically attractive. [DERIVED]


═══════════════════════════════════════════════════════════════
§10. PMNS proof-of-concept and the both-walls theorem
═══════════════════════════════════════════════════════════════

§10.1 Statement of the surviving proof-of-concept

A phenomenological PMNS fit survives in the coalition scan with

    χ²_min = 5.08                                         (10.1)

at

    ε = 11.52,   ρ_e = 0.562,   ρ_ν = 17.23.             (10.2)

This fit is numerically real. It is therefore kept in Paper 3 as a
proof-of-concept datum. [VERIFIED]

However, the same scan systematically tested these fitted numbers against the
catalog of structural constants used elsewhere in the program and found no
clean derivations. The closest curiosities were:

    ε ≈ 2π²/√3,   ε ≈ φ⁵ + φ⁻²,                        (10.3)
    ρ_e ≈ 1/√π,                                          (10.4)
    ρ_ν ≈ 6π − φ.                                        (10.5)

But these are numerological matches, not mechanisms. In particular, the 6π − φ
match is numerically striking yet structurally empty: the integer 6 has no
established role in the D_IV geometry used elsewhere in the construction.
Paper 3 therefore records Eq. (10.2) as phenomenology and rejects
Eqs. (10.3)–(10.5) as derivations. [VERIFIED]

§10.2 Both walls must fall

The coalition analysis also established a stronger structural constraint.

  • Volume disparity alone fails: Δ ≠ 0 with ε = 0 breaks the
    charged-lepton/neutrino degeneracy but collapses the atmospheric angle.
  • Lift alone fails: ε > 0 with Δ = 0 leaves the system overconstrained
    and does not repair the reactor sector.

Therefore quantitative PMNS closure requires both

    ε > 0,   Δ ≠ 0.                                      (10.6)

This is the "both-walls theorem." [VERIFIED]

But §9 immediately sharpens the point: if Δ is restricted to derived boundary
ratios, then it is discrete, not freely adjustable. So inside strict zero-fit
J₃(𝕆ₛ), the both-walls theorem does not open a large two-parameter algebraic
solution space; it exposes the absence of one. [DERIVED]

§10.3 Relation to current neutrino data

Paper 2 (v3.6b) had already moved away from strong claims about quantitative
PMNS angles. Paper 3 makes that retreat formal. The phenomenological fit in
Eq. (10.2) is not the output of the algebra. It is a useful benchmark for what
a successful extension would have to reproduce. This caution is reinforced by
NuFIT 6.0 itself, which reports that in normal ordering the present global fit
remains consistent with CP conservation within 1σ. So even the phase target is
less rigid than older oscillation-data narratives suggested.

§10.4 Status
  • PMNS proof-of-concept fit, Eq. (10.2): [VERIFIED]
  • Derivation of ε, ρ_e, ρ_ν from tested J₃(𝕆ₛ) constants: [DEAD]
  • Quantitative PMNS closure in the strict framework: [OPEN]


═══════════════════════════════════════════════════════════════
§11. Freudenthal triple systems: a bigger house with the same locked door
═══════════════════════════════════════════════════════════════

The natural algebraic extension of J₃(𝕆ₛ) is its Freudenthal triple system

    F(J) = ℝ ⊕ ℝ ⊕ J ⊕ J,                               (11.1)

which for J₃(𝕆ₛ) is 56-dimensional and carries the quartic invariant structure
associated with groups of type E₇.

The coalition's Round 2 verdict was sharp: enlarging to F(J₃(𝕆ₛ)) does not by
itself solve the CKM problem. The reason is conceptual. The FTS enlarges the
representation space and introduces a quartic invariant, but the basic algebraic
machinery remains polynomial in the underlying Jordan data. The
scalar-on-Peirce obstruction is therefore not removed; it is only relocated to
a larger polynomial setting. In that sense the FTS is, algebraically, "a bigger
house with the same locked door." [VERIFIED]

Paper 3 adopts that verdict. At the purely polynomial level:
  • J₃(𝕆ₛ) is too small to yield a single CKM generator.
  • F(J₃(𝕆ₛ)) is larger, but its canonical polynomial operations are still
    not enough to generate the needed asymmetric sector extraction.

The only surviving opening in this direction is not the FTS as such, but the
full E₇-group action, which is non-polynomial when translated back into Jordan
variables. That route is not developed here; it becomes a Paper 5 target.
[OPEN]

A caution is needed on orbit language. Some internal notes attached specific FTS
orbit labels to particular embeddings of the rank-2 vacuum. The precise orbit
assignment depends on how the vacuum is embedded into F(J), and Paper 3 does
not require that detail. Its claim is narrower and robust: FTS enlargement
alone does not evade the polynomial no-go. [DERIVED]


═══════════════════════════════════════════════════════════════
§12. Minimal extensions beyond strict J₃(𝕆ₛ)
═══════════════════════════════════════════════════════════════

The point of a boundary paper is not only to close doors but to identify the
smallest open ones.

Route 1: Sector-dependent vacua Q_u ≠ Q_d
Instead of a single vacuum controlling both up- and down-type flavor sectors,
introduce two nearby vacua, Q_u ≠ Q_d. This is the smallest phenomenological
extension. It immediately makes a true mismatch V_CKM = U_u† U_d possible in
principle. Its cost is equally immediate: it breaks the one-vacuum economy that
was central to Papers 1–2. Unless a relation between Q_u and Q_d is
independently derived, this route is effective but not elegant. [OPEN]

Route 2: FTS/E₇
At the polynomial level, FTS pushes the problem rather than solving it (§11).
But full E₇ group action may act nontrivially enough on the enlarged charge
space to evade the sectorwise-scalar obstruction. That possibility is real and
belongs naturally in Paper 5. Its cost is that the framework stops being a pure
J₃(𝕆ₛ) paper. [OPEN]

Route 3: Non-nuclear vacuum deformation
The non-associator no-go relies crucially on the vacuum being real diagonal,
hence nuclear. A deformed vacuum Q_vac → Q_vac + δQ with non-diagonal or
non-nuclear pieces may evade Theorem 2. This is the most direct way to try to
reawaken associator-based mixing. It is also the most dangerous, because nearly
every success claimed in Papers 1–2 depends on spectral diagonalization in the
fixed Jordan frame. So this route could solve the CKM obstruction only by
sacrificing the algebraic simplicity that produced the earlier relations. It is
mathematically interesting and physically expensive. [OPEN]

Route 4: What does NOT count as an exit
Paper 3 explicitly rejects the following as legitimate exits:
  • reverse-engineered constants such as 6π − φ offered merely because they
    fit a PMNS scan minimum;
  • Berry-phase or monodromy language without a computation;
  • post hoc numerical ratios promoted from coincidence to prediction;
  • hidden extra normalizations masquerading as derived sector data.
Those are not exits; they are rebrandings of fit. [DERIVED]


═══════════════════════════════════════════════════════════════
§13. Methodology, coalition verification, and proposition registry
═══════════════════════════════════════════════════════════════

§13.1 Method

This project uses a declared multi-AI adversarial workflow under PI control.
That methodology is relevant here because Paper 3 is primarily a
theorem-and-audit paper. The decisive results in this draft were not generated
by forward speculation but by repeated hostile checking:
  • ChatGPT / Claude.AI / Cowork converged on the CKM polynomial no-go and
    the non-associator no-go;
  • Claude.AI / Cowork isolated the Jordan-vs-matrix associator split and
    numerically checked it on 500 random Peirce triples;
  • Gemini / Cowork isolated the both-walls PMNS obstruction and confirmed
    that the best-fit PMNS parameters are phenomenological rather than
    structural;
  • Grok / Cowork sharpened the P73 reinterpretation while downgrading
    Berry-phase language to narrative only.

This does not certify the physics. It does something narrower and, for Paper 3,
more valuable: it reduces the chance that the draft is hiding an algebraic
loophole or silently recycling a dead route. [VERIFIED]

§13.2 Proposition registry

The core statements of Paper 3 are:

  1. T1. CKM polynomial no-go — no single generator from the strict
     polynomial J₃(𝕆ₛ) toolkit yields full CKM hierarchy. [DERIVED]
  2. T2. Non-associator structural lemma — real diagonal vacua are nuclear;
     vacuum-inserted associators vanish; cyclic 3-cycle associators split
     into a diagonal-real Jordan piece and an imaginary-diagonal matrix
     piece. [DERIVED]
  3. A-split. Jordan vs matrix associator distinction — only the matrix
     associator exits the Jordan algebra; the Jordan associator remains
     diagonal in J₃(𝕆ₛ). [VERIFIED]
  4. H3/H4. Hua polynomials and kernel-power integrals for D_IV(3), D_IV(4).
     [DERIVED]
  5. R34. Only algebraic Shilov ratio: Ṽ₃/Ṽ₄ = 3/2. [DERIVED]
  6. P73-geo. P73 geometric reinterpretation:
     tan θ_C − sin²θ_W = α/Ṽ₃. Meaningful but unproved. [PLAUSIBLE]
  7. PMNS-fit. Phenomenological PMNS proof-of-concept:
     χ² = 5.08, (ε, ρ_e, ρ_ν) = (11.52, 0.562, 17.23). [VERIFIED]
  8. PMNS-zero-fit. Derivation of those parameters from tested J₃(𝕆ₛ)
     constants. [DEAD]
  9. A-phi. Wolfenstein A = φ⁻¹/² or any tested canonical associator
     variant. [DEAD]
  10. λ-phi. Bare λ = φ⁻³ as exact identity. [DEAD] (~27σ, Gemini R7)
  11. FTS-poly. Polynomial FTS enlargement solves CKM. [DEAD] in the strong
      form "solves," but the weaker statement "FTS pushes the problem without
      solving it" is [VERIFIED].
  12. Berry-CP. CP as Berry phase from G₂₍₂₎ torsion. [CONJECTURAL]

§13.3 Back-feed into Papers 1–2

If Paper 3 is accepted in substance, the companion papers should be updated as
follows.

  • Paper 1 v1.48 → v1.49: Any language implying a single strict-J₃(𝕆ₛ) CKM
    generator should be removed. θ_C formula reconciliation (§6.4) should be
    resolved. P73 should be labeled plausible/geometric.
  • Paper 2 v3.6b → v3.7: The Wolfenstein A route should be archived as dead,
    not left "pending." CKM section should cite Paper 3 no-go theorem.
  • Both papers: The sentence "D_IV(3) and D_IV(4) Hua integrals are unknown"
    should be replaced by "their normalized Hua/Bergman data are explicit, but
    the flavor map is open." PMNS quantitative closures should be labeled
    phenomenological unless and until ε, ρ_e, ρ_ν are derived.

Paper 3 is therefore not just a third manuscript; it is a correction layer for
the whole program. [DERIVED]


═══════════════════════════════════════════════════════════════
§14. Conclusion
═══════════════════════════════════════════════════════════════

Paper 3 proves that the strict single-vacuum polynomial flavor program based on

    Q_vac = φ²e₁ + e₂ ∈ J₃(𝕆ₛ)

has a real boundary, and that this boundary is already reached.

The two central results are negative but clean. First, no single polynomial
construction from the canonical J₃(𝕆ₛ) operations yields a CKM generator with
the observed three-scale hierarchy. Second, the non-associator route fails for
structural reasons: real diagonal vacua are nuclear, so vacuum-inserted
associators vanish; the only nonzero 3-cycle associators are annihilated by the
canonical rank-2 weightings. The program therefore cannot be pushed to full
quark mixing closure merely by trying harder inside the same algebraic box.
[DERIVED]

At the same time, Paper 3 removes one false obstacle. The normalized
Hua/Bergman data for D_IV(3) and D_IV(4) are explicit. What is missing is not
the low-dimensional symmetric-domain mathematics but the mechanism converting
those data into PMNS mismatch parameters. This matters because it replaces a
vague appeal to "unknown integrals" with a sharper statement: the real wall is
the algebra-to-physics map. [DERIVED]

The surviving PMNS proof-of-concept is honest but limited. The best-fit point
(ε, ρ_e, ρ_ν) = (11.52, 0.562, 17.23) is numerically viable, but every
tested attempt to derive those numbers from the framework's structural
constants fails the zero-fit standard. So the proof-of-concept is a target for
future extension, not a closure of the present paper. [VERIFIED]

The practical outcome is that the flavor program now has a clean fork. Either
one accepts a minimal phenomenological extension — distinct sector vacua,
explicit fit parameters, or both — or one leaves strict J₃(𝕆ₛ) and moves into
a larger non-polynomial arena such as full E₇ action. Paper 5 is the place to
test those exits. The role of Paper 3 is to ensure that, whatever comes next,
it starts from explicit no-go theorems rather than from recycled hopes.
[DERIVED]


═══════════════════════════════════════════════════════════════
Appendix A. Detailed proofs for the associator sector
═══════════════════════════════════════════════════════════════

A.1 Nuclearity of real diagonal elements

Let D = diag(d₁, d₂, d₃), d_i ∈ ℝ. Then for any matrices A, B over 𝕆ₛ,

    [(DA)B]_ik = Σ_j (d_i a_ij) b_jk = d_i Σ_j a_ij b_jk = [D(AB)]_ik.

Hence [D, A, B]_mat = 0. Similarly,

    [(AD)B]_ik = Σ_j a_ij (d_j b_jk) = [A(DB)]_ik,

and

    [A(BD)]_ik = [(AB)D]_ik.

So any real diagonal matrix lies in the nucleus of the underlying matrix
algebra. In particular, Q_vac and each e_i are nuclear. [DERIVED]

A.2 Matrix associator on a cyclic Peirce triple

With X = Φ₁₂(a), Y = Φ₂₃(b), Z = Φ₃₁(c), one computes

    XY has ab in position (1,3) and āb̄ in position (3,1).
    YZ has bc̄ in position (1,2) and b̄c in position (2,1).

Hence

    (XY)Z = ((ab)c) · e₁,    X(YZ) = a(bc) · e₁,

and therefore

    [X, Y, Z]_mat = [a, b, c]_oct · e₁.

Since [a, b, c]_oct is purely imaginary, the output is not Hermitian. [DERIVED]

A.3 Jordan associator on a cyclic Peirce triple

The Jordan products are

    X ∘ Y = ½(XY + YX) with ab in position (1,3).
    Y ∘ Z = ½(YZ + ZY) with bc̄ in position (1,2).

A direct computation gives

    (X ∘ Y) ∘ Z = ½ Re((ab)c) · (e₁ + e₃),
    X ∘ (Y ∘ Z) = ½ Re((ab)c) · (e₁ + e₂),

using cyclicity of the real part in an alternative *-algebra. Subtracting,

    [X, Y, Z]_J = ½ Re((ab)c) · (e₃ − e₂).

This is diagonal and real. [DERIVED]

A.4 Canonical vacuum weightings kill the cycle

Under U_Q-type weights,

    w₁₂(U) ~ λ₁λ₂,   w₂₃(U) ~ λ₂λ₃,   w₃₁(U) ~ λ₃λ₁.

At λ₃ = 0, w₂₃(U) = w₃₁(U) = 0. Under Q×·-type weights,

    w₁₂(×) ~ λ₃,   w₂₃(×) ~ λ₁,   w₃₁(×) ~ λ₂,

so w₁₂(×) = 0. Because both associators are trilinear, the weighted 3-cycle
vanishes in either canonical channel. [DERIVED]


═══════════════════════════════════════════════════════════════
Appendix B. Hua-polynomial derivation
═══════════════════════════════════════════════════════════════

For an irreducible bounded symmetric domain of type (r, a, b), the genus is
p = (r−1)a + b + 2. For the type-IV domain D_IV(n), one has
(r, a, b, p) = (2, n−2, 0, n).

The generic Hua polynomial formula is

    χ(λ) = ∏_{j=1}^{r} (λ + 1 + (j−1)a/2)^{1+b+(r−j)a}.

Substituting r = 2, a = n−2, b = 0 gives

    χ_n(λ) = (λ+1)^{n−1} · (λ + n/2).

Hence

    χ₃(λ) = (λ+1)(λ+2)(λ+3/2),
    χ₄(λ) = (λ+1)(λ+2)²(λ+3).

With normalized invariant measure dμ_n,

    ∫_{D_IV(n)} h_n(z,z)^λ dμ_n(z) = χ_n(0)/χ_n(λ).

This yields Eqs. (8.10)–(8.14) in the main text. [DERIVED]


═══════════════════════════════════════════════════════════════
Appendix C. Numerical addendum: PMNS fit and coincidence scan
═══════════════════════════════════════════════════════════════

C.1 PMNS fit summary

Coalition best fit:
    χ²_min = 5.08,   (ε, ρ_e, ρ_ν) = (11.52, 0.562, 17.23).
Status: the fit is real but phenomenological. [VERIFIED]

C.2 Tested structural candidates

    ε ≈ 2π²/√3,   ε ≈ φ⁵ + φ⁻²,
    ρ_e ≈ 1/√π,   ρ_ν ≈ 6π − φ.

Status: numerical curiosities only; no derivation mechanism found.
[DEAD] as zero-fit derivations.

C.3 One-hit pattern scan

    v₄/Ṽ₄ = π²/32 ≈ 0.308

numerically shadows sin²θ₁₂, but

    v₃/Ṽ₃ = π/12 ≈ 0.262   misses sin²θ₁₃,
    v₅/Ṽ₅ = π²/120 ≈ 0.082  misses sin²θ₂₃.

All nine n↔angle assignments were checked; one hit is not a theory. [VERIFIED]


═══════════════════════════════════════════════════════════════
Appendix D. References
═══════════════════════════════════════════════════════════════

[FIX-7: Expanded from stub]

Algebraic-physics program:
  [1] C. Furey, "Standard Model Physics from an Algebra?" PhD thesis,
      University of Waterloo (2015). arXiv:1611.09182
  [2] I. Todorov, S. Drenska, "Octonions, Exceptional Jordan Algebra and
      the Role of the Group F₄ in Particle Physics," Adv. Appl. Clifford
      Algebras 28 (2018) 82. arXiv:1805.06739
  [3] M. Dubois-Violette, I. Todorov, "Exceptional Quantum Geometry and
      Particle Physics II," Nucl. Phys. B 938 (2019) 751. arXiv:1808.08110
  [4] L. Boyle, S. Farnsworth, "The Standard Model, the Pati-Salam Model,
      and 'Jordan Geometry'," New J. Phys. 22 (2020) 073023. arXiv:1910.11888

Bounded symmetric domains and Hua integrals:
  [5] L.K. Hua, "Harmonic Analysis of Functions of Several Complex Variables
      in the Classical Domains," AMS Translations 6 (1963).
  [6] J. Faraut, A. Korányi, "Analysis on Symmetric Cones," Oxford
      University Press (1994). [Ch. XI: Hua integrals]

Freudenthal triple systems and E₇:
  [7] H. Freudenthal, "Beziehungen der E₇ und E₈ zur Oktavenebene, I-XI,"
      Indag. Math. (1954–1963).
  [8] M. Günaydin, "Lectures on Spectrum Generating Symmetries and
      U-Duality," arXiv:0908.0374 (2009).

Jordan algebras:
  [9] K. McCrimmon, "A Taste of Jordan Algebras," Springer (2004).
  [10] I. Yokota, "Exceptional Lie Groups," arXiv:0902.0431 (2009).

Experimental benchmarks:
  [11] Particle Data Group, "Review of Particle Physics," Phys. Rev. D 110
       (2024) 030001. [CKM global fit: λ, A, ρ̄, η̄]
  [12] NuFIT 6.0, http://www.nu-fit.org/ (2024). [PMNS global fit, NO]


═══════════════════════════════════════════════════════════════
END OF PAPER 3 v0.1 DRAFT
═══════════════════════════════════════════════════════════════