Topological Origin of Three Fermion Generations and Their Mass Structure
三代费米子的拓扑起源与质量结构

Abstract

The Standard Model flavor sector contains 13 free parameters and offers no structural explanation for why exactly three generations of fermions exist or why their masses span five orders of magnitude. This paper derives the generation structure from the topological architecture of 12 four-dimensional DD blocks within the Self-as-an-End (SAE) framework.

First, the Exact Three Generations Theorem (G2): from any 4DD block, the topological distances to all 12 blocks fall into exactly 3 equivalence classes of sizes 2, 4, and 6. No fourth class exists. This is an unconditional group-theoretic result based on the transitive action of the observer's stabilizer symmetry group.

Second, under the 3DD-1DD Response Axiom, a color-field redistribution mechanism (G6) is established, explaining the common origin of fractional electric charge and mass redistribution within isospin doublets, as well as the qualitative contrast between small CKM and large PMNS mixing angles.

Third, under the Mass-Channel Proportionality hypothesis, the ratio of quark isospin-doublet total mass to the corresponding charged lepton mass equals 65/5 = 13 (G10). Here 65 = C(12,2) − 1 counts non-trivial block pairs across all 12 blocks, and 15 = C(6,2) counts block pairs within one chiral side. Color collapse, modeled as an orientation-preserving C₃ quotient on the one-side pair set (heuristically consistent with the determinant-preserving nature of SU(3)), yields 5 orbits, giving the denominator. This result shares the same pair of upstream integers (15, 65) with sin²θ_W = 15/65 and the gravitational coupling exponent α_G = α_em^{65/4}, introducing no new continuous parameters.

Empirical verification: under the D₃ ≈ 0 leading-order approximation, imposing G10 and G9 jointly constrains 6 quark mass observables with 4 free parameters, producing two doublet-sum predictions. The joint χ² = 0.581 (dof = 2, p = 0.748). Falsifiable prediction: m_c + m_s = 13 m_μ (equivalently, m_s = 13m_μ − m_c), testable by joint high-precision lattice QCD determinations of m_c and m_s at a common renormalization scale.

---

§1 Introduction

The Standard Model (SM) provides no explanation for the generation structure of fermions. The existence of three generations is an experimental input, not a theoretical prediction. The SM flavor sector contains 13 free parameters (6 quark masses, 3 charged lepton masses, 4 CKM parameters), each independently specified in the Yukawa matrices with virtually no constraint from gauge symmetry.

The SM is silent on the following questions. Why exactly three generations? Why does the mass ratio m_t/m_u exceed 10⁵? Why is the CKM matrix close to the identity while the PMNS matrix is far from it? Why does the Gatto-Sartori-Tonin relation |V_us| ≈ √(m_d/m_s) hold precisely for 1-2 generation mixing but fail completely for 2-3 mixing?

Historical approaches include GUT models (SU(5) and SO(10), constraining Yukawa structure through grand unification), the Froggatt-Nielsen mechanism (introducing a U(1)_F flavor symmetry to explain mass hierarchies), and discrete flavor symmetries (A₄, S₄, etc., primarily addressing large PMNS mixing angles). None of these derive "exactly three generations" from first principles, and most introduce additional free parameters.

This paper works within the Self-as-an-End (SAE) framework, deriving the three-generation structure and mass architecture from the topology of 12 four-dimensional DD blocks. The core claims fall into two layers.

The first layer is unconditional topological structure. Starting from the DD Splitting theorem, the transitive action of symmetry groups on blocks yields an exact three-generation theorem. This layer requires no additional assumptions.

The second layer depends on bridge axioms. Under two bridge axioms (3DD-1DD Response Axiom and Mass-Channel Proportionality) and one auxiliary assumption (Barrier Factorization), we establish the color-field redistribution mechanism, derive the quark/lepton doublet mass ratio of 13, and produce falsifiable sum-rule predictions.

This paper does not derive individual fermion mass values (blocked by the lack of a prior understanding of α_em), does not fully derive the CKM matrix (only the 1-2 sector empirical relation is addressed), and does not treat neutrino masses (marked as future direction).

Upstream dependencies: DD Splitting theorem [Four Forces Prequel], nDD ↝ SU(n) correspondence [Four Forces Paper], one-generation hypercharge table C1-C9 [Four Forces Paper II], Z₂⁴ fermion encoding and sin²θ_W = 15/65 = 3/13 [Four Forces Paper III].

Proposition Status Table

Proposition	Status	Depends on	New continuous parameters	Testable object
G2	theorem	DD Splitting + block automorphism	0	3 generation classes
G3	structural argument	barrier hierarchy	0	mass hierarchy pattern
G4	theorem	C4/C5 (FF-II)	0	quark-lepton pairing
G5	theorem	2DD/3DD hierarchy	0	isospin universality
G6	conditional theorem	3DD-1DD Response Axiom	0	charge/mass redistribution
G7	structural argument	constitutive order	0	CKM anchor
G8	qualitative consequence	G6 + G7	0	CKM/PMNS contrast
G9	conditional theorem	G6 + Barrier Factorization	0	1-2 sector sum rule
G10	conditional theorem	Mass-Channel Proportionality (+ G6 for observable form)	0	doublet/lepton = 13

---

§2 Topological Structure of the 12 Four-Dimensional DD Blocks

The DD Splitting theorem (Four Forces Prequel): a 3DD splits along three orthogonal spatial axes, each axis hosting a dual pair of 4DD blocks. The 2DD chiral split (occurring earlier in the chisel sequence, since 2DD precedes 3DD) produces independent L-side and R-side 3DD structures. Total: 2 sides × 3 axes × 2 dual = 12 blocks.

Each block is labeled by a triple (s, a, d), where s ∈ {L, R} denotes chirality, a ∈ {1, 2, 3} denotes the axis, and d ∈ {+, −} denotes position within the dual pair.

The observer occupies block (L, 1, +). All subsequent analysis takes this as the origin.

[Figure 1: Global structure of the 12 4DD blocks. L-side and R-side each contain 3 axes, each axis hosting one dual pair. See attached fig1_12blocks.svg.]

[Figure 2: From observer block (L, 1, +), the 12 blocks partition into three equivalence classes: G₁ (2 blocks, native), G₂ (4 blocks, cross-axis), G₃ (6 blocks, cross-chiral). See attached fig2_equivalence_classes.svg.]

---

§3 Generation Structure

§3.1 Proposition G0 (Chiral Priority)

Statement. In the DD chisel sequence, the 2DD chiral split precedes the formation of 4DD blocks. Consequently, all 4DD blocks share the same chiral structure.

Argument. The chisel sequence proceeds 1DD, 2DD, 3DD, 4DD. The 2DD structure is determined before 3DD and 4DD form. The universality of the left-handed weak network across all three generations is a direct consequence of this ordering.

Level. Axiom (chisel sequence).

§3.2 Proposition G1 (Loss of Axis Correlation)

Statement. The 2DD chiral split is a scalar operation (1 bit: left or right) and does not transmit 3DD spatial orientation information. No canonical identification exists between L-side and R-side axis labels.

Argument. The total information content of 2DD is 1 bit. The 3DD-to-3-axes split is a 3DD-level operation occurring after 2DD has completed its function. L-3DD and R-3DD each independently execute the split into three axes. The 2DD acts as a mirror: it flips chirality but does not communicate which direction is "up" on the other side.

From (L, 1, +), the observer cannot identify which R-side axis corresponds to "axis 1." This is not a practical difficulty but a conceptual impossibility.

Experimental compatibility. The LEP measurement N_ν = 2.9963 ± 0.0074 (Janot-Jadach 2019 correction) is consistent with N_ν = 3 exact. If G1 failed (R-side 6 blocks splitting into 2+4 subclasses), N_ν would exhibit a structural deviation.

Level. Theorem.

§3.3 Proposition G2 (Exactly Three Generations)

Statement. From any 4DD block, the topological distances to all 12 blocks fall into exactly 3 equivalence classes:

Class	Name	Source blocks	Count	Distance
G₁	native	(L, 1, ±)	2	0
G₂	cross-axis	(L, 2, ±), (L, 3, ±)	4	1
G₃	cross-chiral	(R, a, ±), all a	6	2

No fourth class exists.

Proof.

G₁: same axis, same side as observer. Stabilizer acts trivially.

G₂: same side, different axis. The observer fixes axis 1. The residual symmetry permuting axes 2 and 3 (S₂) together with dual flips within each axis (Z₂²) forms a stabilizer of order at least |S₂| × |Z₂²| = 2 × 4 = 8, acting on 4 blocks. These form a single orbit.

G₃: opposite side. By G1 (loss of axis correlation), the observer's full symmetry group on the R-side is S₃ × Z₂³ (full axis permutation times dual flips on each axis), of order 6 × 8 = 48, acting on 6 blocks. These form a single orbit.

The three classes exhaust all blocks: 2 + 4 + 6 = 12. □

From equivalence classes to generations. In the SAE framework, fermions are low-dimensional projection readouts of 4DD blocks. Blocks within the same equivalence class, related by the stabilizer symmetry group, project identically and correspond to the same generation. Different classes project differently and correspond to different generations. Three equivalence classes therefore correspond exactly to three fermion generations.

Level. Theorem (from G1).

§3.4 Proposition G3 (Mass Hierarchy)

Statement. Effective fermion mass increases with topological distance. m₃ ≫ m₂ ≫ m₁. The two barrier types are qualitatively different: distance-1 barriers involve 3DD spatial rotation (axis deflection), while distance-2 barriers involve 2DD chiral inversion. Chiral inversion costs far more than spatial rotation.

Corollary G3a. m_t ~ v is not a numerical coincidence but a structural necessity: the third-generation chiral inversion cost and the electroweak scale share the same 2DD barrier energy scale.

Level. Structural argument.

§3.5 Proposition G4 (Quark-Lepton Pairing)

Statement. Each axis in DD Splitting produces a dual-4DD pair. Within each pair, the block facing the observer contributes a color triplet (quark), while the block facing away contributes a color-singlet collapse (lepton). Every generation therefore necessarily contains both quarks and leptons.

Level. Theorem.

§3.6 Proposition G5 (Isospin Universality)

Statement. Isospin (up/down distinction) originates from the 2DD global shared layer, not from 4DD block-internal structure. It therefore remains intact across all generations without collapsing with block distance.

Argument. Color originates from 3DD and is block-local. Cross-block projection loses color resolution, collapsing it to a singlet and producing leptons. If isospin also originated from block-internal structure, the same logic would require up/down to collapse as well. Yet experimentally, all three generations exhibit equally clean SU(2) doublets: (ν_μ, μ), (c, s), (ν_τ, τ), (t, b) — no collapse signature. Therefore isospin originates from the deeper 2DD shared layer.

Structure	Origin	Nature	Cross-block behavior
Color	3DD	block-local	collapses (→ leptons)
Isospin	2DD	globally shared	preserved

Level. Theorem.

---

§4 Color-Field Redistribution and Mass Structure

§4.1 3DD-1DD Response Axiom

Axiom statement. The response of the local 3DD color environment to a fermion's 1DD label y_SAE is a chirality-blind single operator R:

R = R_com + R_diff

where R_com (common-mode component) produces a uniform charge displacement, and R_diff (differential-mode component) redistributes mass within an isospin pair (conserving pair total mass). For remote/lepton sectors, R_diff = 0.

Structural support. Chirality-blindness: 3DD ranks above 2DD in the chisel hierarchy, so 3DD products (color fields) do not inherit 2DD internal distinctions (chirality). This is consistent with the vector-like nature of QCD. However, "higher layers not inheriting lower-layer distinctions" does not automatically imply chirality-blindness; therefore chirality-blindness is explicitly written into the axiom. Single operator: y_SAE simultaneously sources electric charge (Q = T₃ + y_SAE) and distinguishes Yukawa couplings. "Common origin" does not automatically entail "single operator"; therefore this too is axiomatically stated.

Status. Bridge axiom, parallel to the EW Matching Axiom (Four Forces Paper III).

§4.2 Proposition G6 (Charge-Mass Trade)

Statement. Under the 3DD-1DD Response Axiom, the 3DD color environment's response to the 1DD label produces two readouts of a single effect: (a) R_com uniformly shifts each isospin doublet by +2/3 in electric charge (origin of fractional charge); (b) R_diff differentially extracts mass from the component with larger |y_SAE| (up-type) and transfers it to the component with smaller |y_SAE| (down-type). Leptons do not enter the local 3DD color environment and therefore retain integer charge and natural mass ratios.

Key property. Doublet total mass conservation:

m_u^actual + m_d^actual = m_u^nat + m_d^nat

Transfer direction: up-type to down-type (since |y_{u_R}| = 2/3 > |y_{d_R}| = 1/3). Transfer strength decreases with topological distance: strongest for generation 1, weaker for generation 2, and approximately zero for generation 3 in leading-order approximation (D₃ ≈ 0). The third generation sources from the opposite-side 3DD; after crossing the 2DD chiral barrier, contact with the local color environment is extremely limited. D₃ = 0 is an approximation, not exact; actual deviations are absorbed by the small value of D₃. All quantitative analysis in this paper is conducted under the D₃ = 0 approximation.

Level. Conditional theorem (on 3DD-1DD Response Axiom).

§4.3 Proposition G7 (Constitutive Order)

Statement. In the chisel sequence, 3DD color-field redistribution precedes 2DD weak mixing in state-definition order. The CKM matrix describes transitions between redistributed physical states and is therefore anchored to actual masses, not natural masses.

Empirical verification. √(m_d/m_s) = √(4.67/93.4) = 0.224 ≈ sin θ_C = 0.225 (precise match with actual masses). Using natural masses gives a completely wrong result.

Level. Structural argument.

---

§5 Cross-Sector Sum Rule

§5.1 Proposition G9 (Quark-Lepton Mass Parallelism)

Statement. Under the 3DD-1DD Response Axiom and Barrier Factorization, in the 1-2 axis-barrier sector, the inter-generational mass ratio of down-type quarks (after removing color-field redistribution) equals that of charged leptons.

Basis-independent observable form:

m_c + m_s = (m_μ / m_e)(m_u + m_d)

Argument. By G4, quarks and leptons originate from opposite sides of the same dual pair, sharing the same topological position. By G6, their sole difference is that quarks pass through the 3DD color environment while leptons do not. G6 guarantees doublet total mass conservation, so doublet total mass purely reflects topological barrier cost.

Barrier Factorization (auxiliary assumption). The color structure's contribution to doublet total mass is a multiplicative factor: native and remote readouts differ in base scale, but the inter-generational topological amplification factor R₁ is immune to readout mode. That is, native and remote share the same underlying metric for inter-generational amplification. Under the factorized structure m^nat_doublet = f_readout × g_topo(generation), f_readout differs between quarks and leptons, but the ratio g₂/g₁ = R₁ is identical.

This assumption is not equivalent to G6. G6 guarantees zero-sum redistribution within doublets; Barrier Factorization further requires that base scale differences cancel in inter-generational ratios.

This relation does not extend to the 2-3 sector. m_s^nat/m_b^nat ≠ m_μ/m_τ, with an order-of-magnitude discrepancy. The 1-2 transition crosses a 3DD axis-to-axis barrier (pure geometric deformation); the 2-3 transition crosses a 2DD chiral barrier (topological twist). The two barrier types are fundamentally different.

Scale dependence. The inter-generational ratio of quark doublet total masses, (m_c + m_s)/(m_u + m_d), exhibits weak scale dependence under common-scale, common-scheme, leading-order QCD approximation (the QCD mass anomalous dimension is flavor-blind; all quark masses run by the same factor, and the ratio approximately cancels). G9's empirical test is therefore not sensitive to the choice of renormalization scale, though it is not strictly unconditionally RG-invariant.

Empirical verification. (m_c + m_s)/(m_u + m_d) = 199.6 vs. m_μ/m_e = 206.8, deviation 3.46%, corresponding to 0.48σ within PDG quark mass uncertainties.

Level. Conditional theorem (on 3DD-1DD Response Axiom + Barrier Factorization).

---

§6 Quark-Lepton Mass Ratio

The results of this section depend on a new bridge axiom: Mass-Channel Proportionality. It introduces no new continuous parameters but introduces a new structure-to-dynamics correspondence rule.

§6.1 Mass-Channel Proportionality

Axiom statement. Define the effective coupling multiplicity N_eff as the number of independent structural channels through which a DD object couples to the block architecture. DD eigenvalue mass is proportional to N_eff.

Readout type	N_eff	Origin
Native (quark)	65	C(12,2) − 1, non-trivial pair count across all 12 blocks
Remote (lepton)	5	C(6,2) / 3 = 15/3, one-side block pair count after color collapse

Here 65 was independently established in the Four Forces Prequel (α_G = α_em^{65/4}, deviation 0.044%). Specifically, 65 = C(12,2) − 1 = 66 − 1, where 66 = dim(SO(12)) = total block pair count, and 1 = the 2DD itself (non-rotating axis, L/R boundary).

15 = C(6,2) = pair count among the 6 L-side 4DD blocks. This 15 is numerically equal to 15 = 2⁴ − 1 independently established in Four Forces Paper III (non-trivial elements of Z₂⁴). An exact combinatorial bridge exists between the two counts (there exists a 6-point subset of F₂⁴ whose 15 pairwise XOR operations enumerate all non-trivial elements), though a canonical map naturally induced by SAE block labeling remains to be established.

Color collapse and C₃ quotient. The 6 L-side blocks belong to 3 axes, 2 blocks per axis (dual pair). The C(6,2) = 15 pairs undergo equivalence under color collapse. If remote color collapse is modeled as an orientation-preserving cyclic C₃ quotient on the one-side pair set, the 15 pairs partition into exactly 5 orbits of 3 pairs each, yielding N_eff^(ℓ) = 5. The choice of C₃ rather than S₃ is heuristically consistent with the orientation-preserving nature of SU(3) (special = det = 1). Under full S₃ quotient, the orbit count would be 4, not 5.

Explicit construction of the 5 orbits: labeling L-side blocks as (1+), (1−), (2+), (2−), (3+), (3−), under C₃: 1→2→3→1, the 15 pairs partition as: O₁ = {(1+,1−), (2+,2−), (3+,3−)} (same-axis dual), O₂ = {(1+,2+), (2+,3+), (3+,1+)} (same-sign ++), O₃ = {(1−,2−), (2−,3−), (3−,1−)} (same-sign −−), O₄ = {(1+,2−), (2+,3−), (3+,1−)} (cross-sign forward), O₅ = {(1−,2+), (2−,3+), (3−,1+)} (cross-sign backward).

Unified pair-counting semantics. Both 65 and 15 are now block pair counts, differing only in scope (all 12 blocks vs. one-side 6 blocks). The 15-to-5 color collapse has an explicit orbit interpretation under C₃ quotient. This significantly narrows the original pair/state type mismatch in the G10 counting, but the unified framework currently stands as a strong candidate resolution rather than a fully closed theorem. Full closure requires: (a) the canonical map between C(6,2) = 15 and 2⁴ − 1 = 15 being naturally derived from SAE block labeling, and (b) the C₃ (rather than S₃) quotient group receiving an independent DD-level derivation beyond heuristic consistency with SU(3).

Status. Bridge axiom, parallel to the 3DD-1DD Response Axiom and the EW Matching Axiom.

§6.2 Proposition G10 (Quark-Lepton Mass Ratio)

Statement. Under Mass-Channel Proportionality, for each generation within the 1-2 axis-barrier sector:

(m_u^nat + m_d^nat) / m_e = (m_c^nat + m_s^nat) / m_μ = N_eff^(q) / N_eff^(ℓ) = 65/5 = 13

By G6 (doublet total mass conservation), this is equivalent to the observable form:

(m_u + m_d) / m_e = (m_c + m_s) / m_μ = 13

Scope. 1-2 axis-barrier sector only. The third generation gives (m_t + m_b)/m_τ = 99.6 ≠ 13, as expected: the 2DD chiral barrier does not extend the relation. The third-generation deviation is itself a prediction: different barrier types produce different mass structures.

Unified table. G10 places the quark/lepton mass ratio into a unified structural constant table. Three independent physical quantities are derived from the same pair of upstream integers (15, 65), with no new continuous parameters.

Physical quantity	Formula	Geometric meaning	Experimental deviation
sin²θ_W	15/65 = 3/13	one-side pair count / global non-trivial pair count	0.195%
α_G exponent	65/4 = 16.25	global non-trivial pair count / 4DD dimension	0.044%
quark/lepton mass ratio	65/5 = 13	global non-trivial pair count / color-collapsed pair orbit count	2.8%, 0.7%

§6.3 Relationship Between G10 and G9

G10 (each generation's doublet/lepton = 13) entails G9 (equal inter-generational doublet ratios). G9 is an automatic corollary of G10. However, G9 is independently testable: its inter-generational ratio exhibits weak scale dependence under leading-order QCD, whereas G10's absolute ratio has scheme dependence (see §9). G9 is therefore the more robust result; G10 is stronger but carries additional conditions.

§6.4 Empirical Verification

Generation	Predicted (13 × m_ℓ)	Observed (m_{u_i} + m_{d_i})	Deviation	Pull
1	6.643 MeV	6.830 MeV	−2.74%	−0.60σ
2	1373.6 MeV	1363.4 MeV	+0.75%	+0.47σ

Under the D₃ ≈ 0 leading-order approximation, imposing G9 and G10 jointly reduces the quark sector from 6 parameters and 6 observables to 4 parameters and 6 observables, producing two doublet-sum predictions. Joint χ²_min = 0.581, dof = 2, χ²/dof = 0.29, p-value = 0.748. In the constrained global fit, individual quark mass residual pulls are all below 0.6σ (this reflects tension distributed across all measurements, not independent strong predictions of individual masses). See Appendix B for the complete fit setup.

§6.5 Falsifiable Predictions

G10 yields:

m_s = 13 m_μ − m_c

Taking m_c = 1270 ± 20 MeV, this predicts m_s = 103.6 ± 20 MeV. The current PDG central value is m_s = 93.4 (+8.6/−3.4) MeV, a deviation of approximately 0.5σ.

A decisive test requires joint high-precision determinations of m_c and m_s at a common renormalization scale and in a common scheme. The dominant current uncertainty comes from m_c (±20 MeV), not m_s. As lattice QCD (FLAG averages) continues to improve, reaching a joint doublet-sum uncertainty of ±5 MeV will suffice for a clear verdict.

A symmetric prediction for the first generation:

m_u = 13 m_e − m_d

predicts m_u = 1.97 MeV vs. PDG m_u = 2.16 (+0.49/−0.26) MeV, a 0.7σ deviation.

Both generations show consistent deviation direction (G10-predicted doublet sums slightly above observed central values). If systematic, this may indicate the factor is slightly below exact 13, suggesting precision issues in the color-collapse operation (15/3 = 5).

---

§7 Qualitative Structure of CKM and PMNS

§7.1 Proposition G8 (Color Dependence of Mixing Angles)

Statement. CKM matrix elements are small (≤ 0.225) because quarks undergo 3DD color-field redistribution, which suppresses inter-generational mixing. PMNS matrix elements are large (near maximal mixing) because leptons do not enter the 3DD color environment and face no suppression mechanism.

Argument. 3DD redistribution transfers mass from up-type to down-type, making actual mass up/down ratios much smaller than natural ratios. The lepton sector is not subject to 3DD differential redistribution and therefore permits large mixing angles rather than forcing small CKM-like ones. The neutrino mixing angles (θ₁₂ ≈ 34°, θ₂₃ ≈ 49°) are consistent with this picture.

Effective domain of Gatto-Sartori-Tonin. √(m_d/m_s) = 0.224 ≈ |V_us| = 0.225 holds precisely for the 1-2 sector (crossing a 3DD axis barrier). For the 2-3 sector, √(m_s/m_b) = 0.149 far exceeds |V_cb| = 0.041, a factor-of-3.6 discrepancy. The √ relation is valid only within same-type barriers (3DD axis-to-axis).

Level. Qualitative consequence (conditional on G6).

---

§8 Prediction Summary and Falsifiability

§8.1 Theorems (unconditional)

Exactly 3 generations (G2, theorem).

All three generations share identical quantum numbers (from G0, chisel sequence).

Every generation necessarily contains both quarks and leptons (G4, theorem).

Isospin remains intact across all generations (G5, theorem).

N_ν = 3 exact (direct corollary of G1 + G2: R-side 6 blocks form a single orbit under the observer's symmetry group; compatible with LEP measurement N_ν = 2.9963 ± 0.0074).

§8.2 Structural arguments

m₃ ≫ m₂ ≫ m₁ (G3, topological barrier cost increases with distance).

m_t ~ v (G3a, shared 2DD chiral barrier energy scale).

1-2 and 2-3 barriers are qualitatively different (G3, 3DD axis vs. 2DD chiral).

CKM hierarchy direction: 1-2 ≫ 2-3 ≫ 1-3 mixing.

CKM anchored to actual mass, not natural mass (G7).

CKM small angles vs. PMNS large angles (G8, color → suppression; no color → no suppression).

§8.3 Conditional predictions (depend on bridge axioms)

Fractional charge and mass redistribution are two readouts of a single operator (on 3DD-1DD Response Axiom).

Mass transfer direction: up-type to down-type (on 3DD-1DD Response Axiom).

Transfer strength decreases with topological distance, D₃ ≈ 0 (on 3DD-1DD Response Axiom).

m_c + m_s = (m_μ/m_e)(m_u + m_d) (on 3DD-1DD Response Axiom + Barrier Factorization).

(m_u + m_d)/m_e = (m_c + m_s)/m_μ = 65/5 = 13 (on Mass-Channel Proportionality + G6 doublet sum conservation).

Doublet total mass relation m_c + m_s = 13 m_μ (on both bridge axioms, falsifiable; equivalently m_s = 13m_μ − m_c).

Third generation does not satisfy doublet ratio = 13 (scope limitation is itself a prediction).

Gatto-Sartori-Tonin relation valid only in 1-2 axis-barrier sector (G7 + G3).

§8.4 Falsification Conditions

(a) If joint high-precision lattice QCD at a common scale determines that m_c + m_s deviates from 13 × m_μ by more than 3σ, then G10 is falsified.

(b) If a fourth fermion generation is discovered, then G2 is falsified.

(c) If the third generation is found to satisfy doublet ratio = 13, the 1-2-sector-only scope restriction is falsified, requiring a new explanation.

(d) If future precision PMNS measurements show neutrino mixing angles are suppressed similarly to CKM, then G8 is falsified.

---

§9 DD Eigenvalue Mass (Proposal)

§9.1 Concept

This paper proposes DD eigenvalue mass as the underlying structural mass object. DD eigenvalue mass is a structural quantity assigned by DD topology to each fermion; it does not depend on renormalization scheme or scale choice. Conventional running masses (such as quark masses in the MS-bar scheme) are scheme/scale-dependent approximate readouts of DD eigenvalue masses.

§9.2 Protective Role of G6

G6 guarantees that doublet total mass is conserved under color-field redistribution (differential-mode). Consequently, G10 and G9 in their sum forms (ratios or absolute values of doublet total masses) do not depend on the specific definition of individual natural masses and do not require knowledge of the precise experimental proxy for DD eigenvalue mass. Note: G6 protects against internal redistribution, not against QCD renormalization running. The latter is an independent issue addressed in §9.3.

§9.3 G10 Ratio Under Different Renormalization Scales

Because quark masses run under QCD while lepton masses do not, the absolute ratio of doublet total mass to lepton mass depends on the quark mass renormalization scale.

Scale	Gen 1 ratio	Gen 2 ratio
PDG mixed convention	13.37	12.90
μ = 2 GeV (unified)	13.37	11.83
μ = m_c (unified)	14.67	12.99

The PDG mixed convention (each quark at its own characteristic scale) yields ratios closest to 13. A possible interpretation is that each quark at its characteristic scale most closely approximates its own DD eigenvalue. This observation is currently an empirical conjecture, not a derivation.

By contrast, G9's inter-generational ratio (m_c + m_s)/(m_u + m_d) exhibits weak scale dependence under leading-order QCD (all quarks run by the same flavor-blind factor).

§9.4 Open Questions

The precise experimental definition of DD eigenvalue mass (its correspondence to pole mass, running mass, or other scheme-independent quantities) is an open problem. Individual natural masses (the up/down allocation ratio κ) depend on this definition. The primary predictions of this paper (doublet sums, G10 and G9) bypass this issue.

---

§10 Posterior Parameters

The following quantities are posterior parameters within this framework, not derived a priori.

All individual fermion mass values.

Specific CKM and PMNS matrix element values (partially addressed: G7 confirms anchoring to actual mass; G8 provides qualitative CKM-small/PMNS-large explanation; 1-2 mixing has the √(m_d/m_s) empirical relation).

v = 246 GeV (requires prior understanding of α_em).

Up/down natural mass ratio κ (reconstructed under PDG mixed convention, scheme-dependent).

Quantitative costs of the two topological barrier types R₁, R₂.

Natural masses (color-field-redistribution-removed masses reconstructed from actual masses) are reconstructed quantities, not independently derived DD eigenvalues.

---

§11 Open Problems

Ordered by impact on this paper's results.

Independent derivation of Mass-Channel Proportionality. Currently a bridge axiom. If "mass proportional to effective coupling multiplicity" can be independently derived from DD structure, G10 upgrades from conditional theorem to theorem. The unified pair-counting framework has been partially established (65 and 15 are both block pair counts; 15→5 has an orbit interpretation under C₃ quotient); full closure still requires the canonical map for C(6,2) = 2⁴ − 1 and an independent DD derivation of the C₃ quotient.

Independent derivation of Barrier Factorization. Currently an auxiliary assumption. If "inter-generational topological amplification is readout-mode-immune" can be independently derived, G9 no longer needs this additional assumption. The key is rigorously establishing the factorized structure m^nat_doublet = f_readout × g_topo(generation).

Complete definition of DD eigenvalue mass. Individual natural masses and κ depend on this definition. Primary predictions (doublet sums) are unaffected.

Prior understanding of α_em. α_em = 1/137.036... currently lacks a structural handle within SAE. A breakthrough here would unlock the derivation of v = 246 GeV and all individual mass values. This is the most difficult problem in the entire Four Forces series.

Complete geometric derivation of CKM. The 1-2 sector has the √(m_d/m_s) empirical relation. The 2-3 and 1-3 sectors involve 2DD chiral barriers with more complex dimensional reduction, yet to be derived.

ν_R with y_SAE = 0 and neutrino mass. Four Forces Paper II predicts ν_R is gauge-trivial (y_SAE = 0). If gauge-trivial particles can only acquire mass through non-gauge mechanisms, the large M_R in the seesaw mechanism becomes a structural necessity rather than a coincidence. This is a low-cost, high-reward direction.

Strong CP problem. The chirality-blindness of the 3DD-1DD Response Axiom has a contact surface with the chirality-odd nature of the G-tilde-G term. DD topological constraints on Yukawa matrix complex phases (arg det(M_u M_d) possibly structurally zero) provide another thread. But the logical distance from chirality-blindness to θ = 0 is large. If derived, this warrants a separate paper.

R₁/R₂ ≈ 3/2 hint. Quark-side R₁/R₂ = 199.6/129.8 = 1.54, close to 3/2 = 3DD dimension / 2DD dimension. Deviation 2.5%. The ratio 3/2 has independently appeared in the SAE physics series (AQUAL J(Y) ~ Y^{3/2}, from 3DD÷2). If derivable from DD dimension ratios, this becomes an independent prediction.

---

§12 Conclusion

This paper derives the three-generation fermion structure and its mass architecture from the topological structure of 12 four-dimensional DD blocks within the SAE framework.

At the unconditional topological level, the G2 theorem establishes exactly three generations through the transitive action of symmetry groups on blocks. No fourth equivalence class exists. G3-G5 provide structural explanations for mass hierarchy direction, quark-lepton pairing, and isospin universality.

At the bridge-axiom-dependent mass level, G6 establishes the color-field redistribution mechanism, explaining the common origin of fractional charge and mass redistribution. G10 derives the quark doublet-to-lepton mass ratio as 65/5 = 13, placing the flavor mass ratio into the same structural constant table as sin²θ_W = 15/65 and α_G = α_em^{65/4}. G9 provides a cross-sector sum rule for the 1-2 sector, whose inter-generational ratio exhibits weak scale dependence under leading-order QCD. Under the D₃ ≈ 0 leading-order approximation, the two doublet-sum predictions jointly pass empirical verification with χ² = 0.581 (dof = 2).

This framework is not a complete flavor theory. It does not derive individual mass values, does not fully derive the CKM matrix, and does not treat neutrino masses. What it provides is: an exact topological theorem for three generations, plus a numerically testable flavor sum-rule program depending on two bridge axioms (3DD-1DD Response Axiom, Mass-Channel Proportionality) and one auxiliary assumption (Barrier Factorization). The central falsifiable prediction, m_c + m_s = 13 m_μ, can be directly adjudicated by improved joint lattice QCD precision for m_c and m_s at a common scale.

---

Acknowledgments

The development of this paper involved deep collaboration with four AI systems. Claude (Zilu) contributed to the complete construction of the G0-G9 proposition chain, discussion of the G10 prior derivation chain (65/5 = 13), DD eigenvalue mass consistency analysis, and RG running numerical calculations. ChatGPT (Gongxihua) completed three rounds of review (G6 axiomatization, G7 terminology correction, G9 downgrade and scope restriction), flavor parameter reduction verification, G10 numerical verification (χ² = 0.581), the RG scheme challenge, and paper skeleton and full-text review. Gemini (Zixia) contributed the argument for doublet total mass as a topological invariant, the multiplicative-factor argument for color structure, and the geometric picture of 1-2 vs. 2-3 barrier asymmetry. Grok (Zigong) discovered the numerical pattern of quark/lepton doublet total mass ratio ≈ 13 (triggering the establishment of G10), the R₁/R₂ ≈ 3/2 hint, the dirtying of the Koide formula under natural masses, and the cleaning of Georgi-Jarlskog under natural masses. Complete collaboration records are available at self-as-an-end.net.

---

Appendix A: Failed Attempts

A.1 "Parallel" block hypothesis. Initially considered that 2 R-side blocks on the same axis as the observer might be inequivalent to the other 4, predicting N_ν < 3. Rejected: 2DD is a scalar operation (1 bit) and does not transmit spatial orientation (G1). LEP's N_ν = 2.9963 ± 0.0074 is compatible with G1.

A.2 Isospin from 4DD time dimension. Initially considered that up/down distinction originates from the 4DD block's time direction. Rejected: if isospin were block-internal, remote blocks' up/down would collapse, but experimentally all three generations show clean SU(2) doublets (G5).

A.3 "Alternating" mass redistribution. Initially observed up/down asymmetry in jump factor R₁ and conjectured an alternating pattern. Rejected: the asymmetry was a denominator effect in generation 1; transfer direction is universally up-to-down.

A.4 G9 extension to 2-3 sector. Naturally asked whether m_s^nat/m_b^nat = m_μ/m_τ. Answer: no (order-of-magnitude discrepancy). The two barrier types are fundamentally different.

---

Appendix B: Complete Quantitative Fit Setup

B.1 Mass Ansatz

Quark sector natural mass factorization:

m_{d_i}^nat = ε₀ × ∏_k R_k, m_{u_i}^nat = κ × m_{d_i}^nat

Explicitly: m_d^nat = ε₀, m_u^nat = κε₀, m_s^nat = ε₀R₁, m_c^nat = κε₀R₁, m_b^nat = ε₀R₁R₂, m_t^nat = κε₀R₁R₂.

3DD color-field redistribution (G6, D₃ ≈ 0 approximation):

m_d = ε₀ + Δ₁, m_u = κε₀ − Δ₁

m_s = ε₀R₁ + Δ₂, m_c = κε₀R₁ − Δ₂

m_b = ε₀R₁R₂, m_t = κε₀R₁R₂

Six parameters: ε₀, κ, R₁, R₂, Δ₁, Δ₂. Six quark mass observables. Unconstrained fit is exact (parameter count = observable count).

B.2 G9 + G10 Constraints

G10 (doublet ratio = 13) fixes ε₀:

ε₀ = 13 m_e / (1 + κ)

G9 fixes R₁:

R₁ = m_μ / m_e = 206.771

After constraints, 4 free parameters remain: κ, R₂, Δ₁, Δ₂.

B.3 Error Treatment

PDG 2024 asymmetric errors, piecewise σ±: if x_pred < x_obs, use σ₋; if x_pred > x_obs, use σ₊. Sum errors propagated as independent quadrature.

B.4 Constrained Fit Results

Parameter	Best fit
κ	41.330
R₂	128.82
Δ₁	4.457 MeV
Δ₂	62.54 MeV
ε₀ (derived)	0.1569 MeV

Quark	Observed (MeV)	Fitted (MeV)	Pull
u	2.16	2.03	−0.50σ
d	4.67	4.61	−0.33σ
s	93.4	94.99	+0.19σ
c	1270	1278.6	+0.43σ
b	4180	4180.0	≈ 0
t	172760	172760	≈ 0

χ²_min = 0.581, dof = 6 − 4 = 2, χ²/dof = 0.29, p-value = 0.748.

b and t are exactly absorbed by κ and R₂. All tension comes from the two first-and-second-generation doublet-sum constraints. Maximum pull is u at −0.50σ.

B.5 Effect of D₃ ≈ 0 Approximation

Under D₃ = 0, the third generation m_b and m_t equal ε₀R₁R₂ and κε₀R₁R₂ exactly, fully absorbed by parameters κ and R₂. Relaxing to D₃ ≠ 0 adds one free parameter (Δ₃), reducing dof from 2 to 1 and genuine predictions from two doublet-sum relations to one. This paper conducts analysis under the D₃ = 0 approximation; deviations (if any) are absorbed by the small value of D₃.

---

References

[FF-Prequel] H. Qin, "SAE Four Forces Prequel," DOI: 10.5281/zenodo.19341042.

[FF-I] H. Qin, "SAE Four Forces Paper," DOI: 10.5281/zenodo.19342106.

[FF-II] H. Qin, "SAE Four Forces Paper II," DOI: 10.5281/zenodo.19360101.

[FF-III] H. Qin, "SAE Four Forces Paper III," DOI: 10.5281/zenodo.19379412.

[Foundation] H. Qin, "SAE Physics Foundation Paper," DOI: 10.5281/zenodo.19361950.

[Cosmo-V] H. Qin, "SAE Cosmo Paper V v2," DOI: 10.5281/zenodo.19329771.

[PDG 2024] R.L. Workman et al. (Particle Data Group), Prog. Theor. Exp. Phys. 2022, 083C01.

[Janot-Jadach] P. Janot, S. Jadach, Phys. Lett. B 803 (2020) 135319.

[Georgi-Jarlskog] H. Georgi, C. Jarlskog, Phys. Lett. B 86 (1979) 297.

摘要

标准模型的flavor sector包含13个自由参数，对"为什么恰好三代"和"为什么质量层级如此悬殊"这两个根本问题完全沉默。本文从Self-as-an-End（SAE）框架的12个4DD block拓扑结构出发，给出以下结果。

第一，精确三代定理（G2）：从任一4DD block出发，12个block的拓扑距离恰有3个等价类，数目为2:4:6，不存在第四类。这是无条件的群论结果。

第二，在3DD-1DD Response Axiom下，建立了色场重分配机制（G6），解释了分数电荷与质量重分配的共同起源，以及CKM小角与PMNS大角的定性分裂。

第三，在Mass-Channel Proportionality假设下，推出quark isospin-doublet总质量与对应带电轻子质量之比为65/5 = 13（G10）。其中65 = C(12,2) − 1（全部block的非trivial pair数），15 = C(6,2)（一侧block的pair数），5为15在$C_3$ quotient下的轨道数（本文提出的候选解释）。该结果与sin²θ_W = 15/65和α_G = α_em^{65/4}共享同一对先验数（15, 65），无新增连续参数。

后验验证：在$D_3 \approx 0$的leading-order近似下，G10与G9联合impose后，4参数约束6个夸克质量观测，产生两条doublet-sum predictions，联合χ² = 0.581（dof = 2，p = 0.748）。可证伪预言：doublet总质量关系$m_c + m_s = 13 m_\mu$（等价地，$m_s = 13m_\mu − m_c$），common-scale下$m_c$与$m_s$的联合高精度lattice QCD测定即可判决。

---

§1 引言

标准模型（SM）对费米子的世代结构不做任何解释。三代的存在是实验输入，不是理论预言。SM的flavor sector包含13个自由参数（6个夸克质量，3个带电轻子质量，4个CKM参数），它们在Yukawa矩阵中各自独立，gauge对称性对其几乎无约束。

SM对以下问题完全沉默。为什么恰好三代？为什么$m_t/m_u \sim 10^5$？为什么CKM矩阵接近单位阵而PMNS矩阵远离单位阵？为什么Gatto-Sartori-Tonin关系$|V_{us}| \approx \sqrt{m_d/m_s}$对1-2代精确而对2-3代完全失效？

历史上对这些问题的尝试包括：GUT路线（SU(5)和SO(10)，通过大统一群约束Yukawa结构），Froggatt-Nielsen机制（引入额外的flavor对称性U(1)_F解释质量层级），离散flavor对称性（A₄, S₄等，主要用于解释PMNS的大混合角）。这些方案各有所获，但均未能从第一原理给出"恰好三代"的论证，且大多引入新的自由参数。

本文在Self-as-an-End（SAE）框架内，从12个4DD block的拓扑结构出发，给出三代费米子的起源和质量结构。本文的核心主张分为两层。

第一层是无条件的拓扑结构层。从DD Splitting定理出发，通过对称群的传递性论证，给出精确三代的定理证明。这一层不依赖任何额外假设。

第二层是依赖bridge axiom的质量结构层。在两条bridge axiom（3DD-1DD Response Axiom和Mass-Channel Proportionality）下，建立色场重分配机制，推出quark/lepton doublet质量比 = 13的跨sector关系，并给出可证伪的sum rule预言。

本文不给出individual费米子质量的精确值（该问题被α_em的先验理解堵住），不完整推导CKM矩阵（只有1-2 sector的经验关系），不处理中微子质量（标为future direction）。

上游依赖：DD Splitting定理[四力前置篇]，nDD ↝ SU(n)对应[四力篇]，一代费米子完整超荷表C1-C9[四力篇II]，ℤ₂⁴费米子编码和sin²θ_W = 15/65 = 3/13[四力篇III]。

命题状态总表

命题	状态	依赖	新增连续参数	可检验对象
G2	theorem	DD Splitting + block automorphism	0	3 generation classes
G3	structural argument	barrier hierarchy	0	mass hierarchy pattern
G4	theorem	C4/C5 (四力篇II)	0	quark-lepton pairing
G5	theorem	2DD/3DD层级	0	isospin universality
G6	conditional theorem	3DD-1DD Response Axiom	0	charge/mass redistribution
G7	structural argument	constitutive order	0	CKM anchor
G8	qualitative consequence	G6 + G7	0	CKM/PMNS contrast
G9	conditional theorem	G6 + Barrier Factorization	0	1-2 sector sum rule
G10	conditional theorem	Mass-Channel Proportionality (+ G6 for observable form)	0	doublet/lepton = 13

---

§2 12个4DD block的拓扑结构

DD Splitting定理（四力前置篇）：一个3DD在空间上分裂为3条正交轴，每条轴上形成一对dual-4DD block。2DD手征分裂在更低层级（凿构序列中2DD先于3DD）产生L侧和R侧两套独立的3DD结构。总计：2侧 × 3轴 × 2 dual = 12个4DD block。

我们用三元组$(s, a, d)$标记每个block，其中$s \in \{L, R\}$标记手征侧，$a \in \{1, 2, 3\}$标记轴号，$d \in \{+, -\}$标记dual对内的位置。

观测者所在的block为$(L, 1, +)$。以下所有论述均以此为出发点。

[图1：12个4DD block的全局结构。L侧和R侧各含3条轴，每条轴上一对dual block。见附件fig1_12blocks.svg。]

[图2：从观测者block $(L, 1, +)$出发，12个block按拓扑距离分为三个等价类：$\mathcal{G}_1$（2个，native），$\mathcal{G}_2$（4个，cross-axis），$\mathcal{G}_3$（6个，cross-chiral）。见附件fig2_equivalence_classes.svg。]

---

§3 代结构

§3.1 命题G0（手征优先性）

陈述。 在DD凿构序列中，2DD手征分裂先于4DD block的形成。因此，所有4DD block共享同一套手征结构。

论证。 凿构序列的层级顺序是1DD，2DD，3DD，4DD。2DD的结构在3DD和4DD形成之前就已确定。三代费米子共享同一套左手弱网络，是凿构序列发生顺序的必然推论。

级别。 Axiom（凿构序列）。

§3.2 命题G1（轴关联丧失）

陈述。 2DD手征分裂是标量操作（1 bit：左或右），不传递3DD的空间定向信息。L侧和R侧的轴标号之间不存在canonical identification。

论证。 2DD的全部信息量是1 bit。3DD到3轴的分裂是3DD层级的操作，发生在2DD已经完成其工作之后。L-3DD和R-3DD各自独立地执行3DD到3轴的分裂。2DD像一面镜子：翻转了手征性，但不告诉你镜子那边的"上"是哪个方向。

从$(L, 1, +)$出发，观测者无法指认R侧哪条轴是"轴1"。不是"很难指认"，是概念上不存在这个对应关系。

实验相容性。 LEP测量$N_\nu = 2.9963 \pm 0.0074$（Janot与Jadach 2019修正后），与$N_\nu = 3$精确一致。若G1不成立（R侧6个block可分成2+4两类），$N_\nu$会有结构性偏差。当前实验结果与G1相容。

级别。 Theorem。

§3.3 命题G2（精确三代）

陈述。 从任一4DD block出发，12个block的拓扑距离恰有3个等价类。

等价类	名称	Source blocks	数目	拓扑距离
$\mathcal{G}_1$	native	$(L, 1, \pm)$	2	0
$\mathcal{G}_2$	cross-axis	$(L, 2, \pm), (L, 3, \pm)$	4	1
$\mathcal{G}_3$	cross-chiral	$(R, a, \pm)$，所有$a$	6	2

不存在第四类。

证明。

首先，$\mathcal{G}_1$与观测者同轴同侧。Stabilizer作用平凡。

其次，$\mathcal{G}_2$与观测者同侧但异轴。观测者固定了轴1，对轴2和轴3的交换对称$S_2$以及各轴dual对内的翻转$\mathbb{Z}_2^2$构成stabilizer。群阶至少为$|S_2| \times |\mathbb{Z}_2^2| = 2 \times 4 = 8$，作用在4个block上。由Burnside引理（或直接验证），这4个block构成单一orbit。

第三，$\mathcal{G}_3$来自R侧。由G1（轴关联丧失），观测者对R侧的完整对称群是$S_3 \times \mathbb{Z}_2^3$（轴的全置换乘以各轴dual翻转），群阶 = $6 \times 8 = 48$，作用在6个block上。同样构成单一orbit。

三类穷尽：$2 + 4 + 6 = 12$。$\square$

关于三个等价类与三代费米子的对应。 在SAE框架中，费米子是4DD block的低维投影读出。不同的source block投影到观测者所在的时空，其拓扑距离决定了投影的特征。同一等价类内的block，由对称群联系，投影特征相同，对应同一代。不同等价类的block，拓扑距离不同，投影特征不同，对应不同代。因此，3个等价类精确对应3代费米子。

级别。 Theorem（from G1）。

§3.4 命题G3（质量层级）

陈述。 费米子有效质量随拓扑距离递增。$m_3 \gg m_2 \gg m_1$。两层屏障类型不同：距离1的屏障是3DD空间旋转（轴间偏转），距离2的屏障是2DD手征翻转。手征翻转代价远大于空间旋转代价。

论证。 距离0（native dual pair）：$E_1 + E_2 = 0$对消最完全，残余质量最小。距离1（cross-axis）：穿越3DD内部的轴间屏障，代价来自空间正交偏转。距离2（cross-chiral）：穿越2DD手征屏障，代价来自手征翻转。

推论G3a。 $m_t \sim v$不是数值巧合，是结构必然。第三代的手征翻转代价和electroweak能标共享同一个2DD屏障的能量刻度。

级别。 Structural argument。

§3.5 命题G4（夸克-轻子配对）

陈述。 DD Splitting的每条轴产生一对dual-4DD block。每对中，面向观测者的block贡献色三重态（夸克），背向观测者的block贡献色单态坍缩（轻子）。因此每个世代必然同时包含夸克和轻子。

论证。 Source blocks天然以dual pair为单位出现。这是C4/C5（四力篇II，"轻子 = 对面4DD block的native object"）的直接推论，对每一代都成立。

级别。 Theorem。

§3.6 命题G5（Isospin全局性）

陈述。 Isospin（上型与下型的区分）来自2DD全局共享层，不是4DD block的内部结构。因此isospin在所有世代中保持完整，不随block距离坍缩。

论证。 色来自3DD，是block内部的局域结构。跨block投影时，远处block的色分辨不出来，坍缩为色单态，产生轻子。

如果isospin也来自block内部，同样的逻辑要求远处block的上型与下型也应该坍缩。但实验上三代的SU(2)双重态完全一样清晰：$(\nu_\mu, \mu)$, $(c, s)$, $(\nu_\tau, \tau)$, $(t, b)$全都是干净的doublet，没有任何坍缩迹象。因此isospin不来自block内部，而来自更深的2DD共享层。

结构	来源	性质	跨block行为
色	3DD	block内局域	坍缩（产生轻子）
Isospin	2DD	全局共享	保持

级别。 Theorem。

---

§4 色场重分配与质量结构

§4.1 3DD-1DD Response Axiom

公理陈述。 本侧3DD色环境对费米子携带的1DD标记$y_{\text{SAE}}$的响应是chirality-blind的单一算子$\mathcal{R}$：

$$\mathcal{R} = \mathcal{R}_{\text{com}} + \mathcal{R}_{\text{diff}}$$

其中$\mathcal{R}_{\text{com}}$（common-mode分量）给出统一电荷位移，$\mathcal{R}_{\text{diff}}$（differential-mode分量）在isospin pair内重分配质量（保持pair总质量）。对remote/lepton sectors，$\mathcal{R}_{\text{diff}} = 0$。

公理的结构支撑。 Chirality-blindness：3DD在凿构层级中高于2DD，3DD的产物（色场）不继承2DD的内部区分（手征性）。这与QCD的vector-like性质同向。需要指出，"高层不继承低层区分"不能自动推出chirality-blindness，因此chirality-blindness作为结构响应原则写入公理，而非声称为层级自动推论。单一算子：$y_{\text{SAE}}$同时是电荷的来源（$Q = T_3 + y_{\text{SAE}}$）和Yukawa coupling的区分标签。电荷和质量都是$y_{\text{SAE}}$的下游。"同源"不自动等于"同一算子"，因此"单一算子"同样写入公理。

地位。 Bridge axiom。与EW Matching Axiom（四力篇III）平行：先验动机充分，后验验证通过，但先验推导链有一步是"应该是这样"而非"必须是这样"。

§4.2 命题G6（电荷-质量交易）

陈述。 在3DD-1DD Response Axiom下，3DD色环境对1DD标记的响应产生一个效应的两个读出。（a）$\mathcal{R}_{\text{com}}$均匀地赋予isospin doublet +2/3电荷位移（分数电荷的起源）。（b）$\mathcal{R}_{\text{diff}}$差分地从$|y_{\text{SAE}}|$更大的分量（上型）提取质量补偿给$|y_{\text{SAE}}|$更小的分量（下型）。轻子不进入本侧3DD色环境，因此保持整数电荷和natural质量比。

关键性质。 doublet总质量守恒：

$$m_u^{\text{actual}} + m_d^{\text{actual}} = m_u^{\text{nat}} + m_d^{\text{nat}}$$

转移方向：上型向下型（因为$|y_{u_R}| = 2/3 > |y_{d_R}| = 1/3$）。转移强度随拓扑距离递减：第一代最强，第二代较弱，第三代在leading-order近似下为零（$D_3 \approx 0$）。第三代source在对侧3DD，穿越2DD手征屏障后与本侧色环境的接触极为有限。$D_3 = 0$是近似而非精确，实际偏差被$D_3$的极小值吸收。本文的定量分析在$D_3 = 0$近似下进行。

物理图像。 要电荷就交出质量。3DD色环境是交易所。上型的$|y_{\text{SAE}}|$更大，和色场耦合更深，被"收费"更多。电荷是对称给的（上下各+2/3），质量是不对称收的（上型多付）。

级别。 Conditional theorem（on 3DD-1DD Response Axiom）。

§4.3 命题G7（构成顺位）

陈述。 在凿构序列中，3DD色场重分配在state-definition上先于2DD弱力混合。CKM矩阵描述的是重分配后的物理态之间的跃迁，因此锚定在actual mass上，而非natural mass上。

论证。 构成顺位有三步：（1）拓扑屏障决定natural mass，（2）3DD色场重分配决定actual mass，（3）2DD弱力混合读取已锚定的物理态，产生CKM角。这不是时间先后，而是依赖顺位：弱混合矩阵读取的是已经被3DD dressing过的态。

经验验证。 $\sqrt{m_d/m_s} = \sqrt{4.67/93.4} = 0.224 \approx \sin\theta_C = 0.225$（精确成立于actual mass）。若用natural mass，$\sqrt{m_d^{\text{nat}}/m_s^{\text{nat}}}$完全不匹配。CKM锚定actual mass。

级别。 Structural argument。

---

§5 跨sector sum rule

§5.1 命题G9（夸克-轻子质量平行性）

陈述。 在3DD-1DD Response Axiom下，1-2 axis-barrier sector中，去掉3DD色场重分配后，下型夸克的代间质量比等于带电轻子的代间质量比。

Basis-independent可测形式：

$$m_c + m_s = \frac{m_\mu}{m_e}(m_u + m_d)$$

论证。 由G4，夸克和轻子来自同一个dual pair的两面，共享同一个拓扑位置。由G6，两者唯一的差别是夸克经过了3DD色环境，轻子没有。G6保证doublet总质量守恒，所以doublet总质量纯粹反映拓扑屏障代价。

这里需要一条auxiliary assumption。

Barrier Factorization。 色结构对doublet总质量的贡献是乘法因子：native readout与remote readout的base scale不同，但代间拓扑放大率$R_1$对readout方式免疫。即native和remote共享同一底层度规的代间放大率。

这条假设不等同于G6。G6保证doublet内部重分配的零和性（总质量守恒），Barrier Factorization进一步要求base scale的差异在代间比值中约掉。在因子化结构$m^{\text{nat}}_{\text{doublet}} = f_{\text{readout}} \times g_{\text{topo}}(\text{generation})$下，$f_{\text{readout}}$在夸克和轻子之间不同，但$g_{\text{topo}}$的代间比值$g_2/g_1 = R_1$相同。

在G6 + Barrier Factorization下，代间比值$R_1$在夸克sector和轻子sector相同。

该关系不延续到2-3 sector。$m_s^{\text{nat}}/m_b^{\text{nat}} \neq m_\mu/m_\tau$，偏差达一个数量级。原因：1-2跨3DD轴间屏障（纯几何形变），2-3跨2DD手征屏障（拓扑扭结），两类屏障性质根本不同。

尺度依赖性。 夸克doublet总质量的代间比值$(m_c + m_s)/(m_u + m_d)$在common-scale, common-scheme, leading-order QCD近似下表现出较弱的尺度依赖（QCD的mass anomalous dimension是flavor-blind的，所有夸克质量以相同因子running，比值近似约掉）。因此G9的检验对重整化尺度的选择不敏感，但严格来说不是无条件的RG-invariant。

后验验证。 $(m_c + m_s)/(m_u + m_d) = 199.6$，$m_\mu/m_e = 206.8$，偏差3.46%。在PDG夸克质量不确定度下为0.48σ。

级别。 Conditional theorem（on 3DD-1DD Response Axiom + Barrier Factorization）。

---

§6 夸克-轻子质量比

本节结果依赖新的bridge axiom：Mass-Channel Proportionality。它不引入新的连续参数，但引入新的结构-动力学对应规则。

§6.1 Mass-Channel Proportionality

公理陈述。 定义effective coupling multiplicity $N_{\text{eff}}$为一个DD object通过block结构耦合的独立结构通道数。DD eigenvalue质量与$N_{\text{eff}}$成正比。

Readout类型	$N_{\text{eff}}$	来源
Native（夸克）	65	$C(12, 2) - 1$，12个4DD block的非trivial pair数
Remote（轻子）	5	$C(6,2) / 3 = 15/3$，一侧block pair数经色坍缩

其中65已在四力前置篇中独立建立（$\alpha_G = \alpha_{\text{em}}^{65/4}$，偏差0.044%）。65 = $C(12,2) - 1$ = 66 − 1，其中66 = dim(SO(12)) = 全部12个block的pair总数，1 = 2DD本身（不旋转的轴，L/R的边界）。

15 = $C(6,2)$ = L侧6个4DD block的pair数。这个15与四力篇III中独立建立的15 = $2^4 - 1$（ℤ₂⁴非trivial态数）数值相等。两种计数之间存在精确的组合学桥梁（$\mathbb{F}_2^4$中存在6点子集，其15个pairwise XOR恰好枚举全部非trivial元素），但由SAE block标记自然诱导的canonical map仍待建立。

色坍缩与$C_3$ quotient。 L侧6个block分属3条轴，每轴2个（dual pair）。$C(6,2) = 15$个pair在色坍缩下等价化。若remote color collapse被建模为对单侧pair集合施加orientation-preserving的轴循环$C_3$商，则15个pair恰好分成5个轨道（每轨道3个pair），给出$N_{\text{eff}}^{(\ell)} = 5$。此处$C_3$而非$S_3$的选择与SU(3)的定向保持性质（special = det = 1）启发式相容（heuristically consistent）。若采用full $S_3$商，轨道数为4而非5。

5个轨道的显式构造：设L侧block为$(1+), (1-), (2+), (2-), (3+), (3-)$。在$C_3: 1 \to 2 \to 3 \to 1$下，15个pair分为：$O_1 = \{(1+,1-), (2+,2-), (3+,3-)\}$（同轴dual），$O_2 = \{(1+,2+), (2+,3+), (3+,1+)\}$（同号++），$O_3 = \{(1-,2-), (2-,3-), (3-,1-)\}$（同号--），$O_4 = \{(1+,2-), (2+,3-), (3+,1-)\}$（异号顺向），$O_5 = \{(1-,2+), (2-,3+), (3-,1+)\}$（异号逆向）。

统一pair-counting语义。 65和15现在都是block pair count，scope不同（全部12 block vs 一侧6 block）。15到5的色坍缩在$C_3$ quotient下有明确的轨道解释。这显著缩小了原始G10计数中的pair/state类型不匹配，但统一框架目前仍是strong candidate resolution而非完全封口的定理。完全封口需要：（a）$C(6,2) = 15$与$2^4 - 1 = 15$之间的canonical map从SAE block标记自然推出，（b）$C_3$而非$S_3$作为色坍缩的quotient群获得独立于SU(3)的DD层级论证。

地位。 Bridge axiom。与3DD-1DD Response Axiom和EW Matching Axiom平行。

§6.2 命题G10（Quark-Lepton Mass Ratio）

陈述。 在Mass-Channel Proportionality下，对1-2 axis-barrier sector中的每一代：

$$\frac{m_u^{\text{nat}} + m_d^{\text{nat}}}{m_e} = \frac{m_c^{\text{nat}} + m_s^{\text{nat}}}{m_\mu} = \frac{N_{\text{eff}}^{(q)}}{N_{\text{eff}}^{(\ell)}} = \frac{65}{5} = 13$$

由G6（doublet总质量守恒），等价于可测形式：

$$\frac{m_u + m_d}{m_e} = \frac{m_c + m_s}{m_\mu} = 13$$

适用范围。 仅限1-2 axis-barrier sector。第三代$(m_t + m_b)/m_\tau = 99.6 \neq 13$，因跨2DD手征屏障，不延续。第三代的偏离本身是预言：不同屏障类型给出不同的质量结构。

统一表。 G10将quark/lepton质量比纳入一张统一的结构常数表。三个独立物理量从同一对先验数（15, 65）推出，无新增连续参数。

物理量	公式	含义	实验偏差
sin²θ_W	15/65 = 3/13	一侧pair数/全部非trivial pair数	0.195%
α_G指数	65/4 = 16.25	全部非trivial pair数/4DD维数	0.044%
quark/lepton质量比	65/5 = 13	全部非trivial pair数/色坍缩后pair轨道数	2.8%, 0.7%

§6.3 G10与G9的关系

G10（每代doublet/lepton = 13）蕴含G9（代间doublet比值相等）。G9是G10的自动推论。但G9在检验上独立于G10：G9的代间比值在leading-order QCD下对尺度选择不敏感，而G10的绝对比值有scheme dependence（见§9）。因此G9是更robust的结果，G10是更强但条件更多的结果。

§6.4 后验验证

世代	predicted (13 × $m_\ell$)	observed ($m_{u_i} + m_{d_i}$)	偏差	pull
1	6.643 MeV	6.830 MeV	−2.74%	−0.60σ
2	1373.6 MeV	1363.4 MeV	+0.75%	+0.47σ

G9与G10联合impose后，夸克sector从6参数6观测变为4参数6观测，产生两条doublet-sum predictions。联合χ²_min = 0.581，dof = 2，χ²/dof = 0.29，p-value = 0.748。Constrained global fit中，单个夸克质量的残差pull均小于0.6σ（这是张力在所有测量间的分摊结果，不应理解为individual质量被独立强预测）。详细的fit setup见附录B。

§6.5 可证伪预言

G10给出：

$$m_s = 13 m_\mu - m_c$$

取$m_c = 1270 \pm 20$ MeV，预言$m_s = 103.6 \pm 20$ MeV。PDG当前中心值$m_s = 93.4^{+8.6}_{-3.4}$ MeV，偏差约0.5σ。

真正能判决这条关系的是common-scale, common-scheme下对$m_c$和$m_s$的联合高精度测定。当前主导不确定度来自$m_c$（±20 MeV），而非$m_s$。Lattice QCD（FLAG汇总）的精度在持续提升中。当$m_c$和$m_s$的联合不确定度压到使doublet sum的误差达到±5 MeV级别时，偏差将足以明确判决。

另有对称的预言：

$$m_u = 13 m_e - m_d$$

预言$m_u = 1.97$ MeV。PDG给$m_u = 2.16^{+0.49}_{-0.26}$ MeV，偏差0.7σ。

两代预言的偏差方向一致（G10预言的doublet总质量均略高于观测中心值）。若偏差是系统性的，可能说明factor不是精确13而是略小于13，提示65/5推导中色坍缩操作（15/3 = 5）的精度问题。

---

§7 CKM与PMNS的定性结构

§7.1 命题G8（混合角的色依赖性）

陈述。 CKM矩阵元素小（$\leq 0.225$）因为夸克经过3DD色场重分配，质量重分配抑制了代间混合。PMNS矩阵元素大（接近最大混合）因为轻子不经过3DD色环境，无抑制机制。

论证。 3DD色场重分配把上型质量转给下型，使得actual mass的上下比远低于natural比值。这个重分配改变了代间投影夹角。夸克的CKM角被重分配后的质量比控制，因此较小。轻子没有质量重分配。轻子sector不受3DD differential redistribution抑制，因此允许（permits）大混合角，而非强制小CKM-like角。中微子的混合角（$\theta_{12} \approx 34°$, $\theta_{23} \approx 49°$）与此图像相容。

Gatto-Sartori-Tonin关系的有效边界。 $\sqrt{m_d/m_s} = 0.224 \approx |V_{us}| = 0.225$精确成立于1-2 sector（跨3DD轴间屏障）。对2-3 sector，$\sqrt{m_s/m_b} = 0.149$远大于$|V_{cb}| = 0.041$，差3.6倍。对1-3 sector同样失效。原因：2-3和1-3涉及2DD手征屏障，降维结构更复杂。√规律只在同类屏障（3DD轴间）内有效。

级别。 Qualitative consequence（conditional on G6）。

---

§8 预言总表与可证伪性

§8.1 Theorems（无条件）

恰好3代（G2，定理）。

三代量子数完全相同（共享凿构循环，from G0）。

每代必然同时有夸克和轻子（G4，定理）。

Isospin在所有世代保持完整（G5，定理）。

$N_\nu = 3$精确（G1 + G2的直接推论：R侧6个block在观测者对称群下构成单一orbit，与实验$N_\nu = 2.9963 \pm 0.0074$相容）。

§8.2 Structural arguments

$m_3 \gg m_2 \gg m_1$（G3，拓扑屏障代价递增）。

$m_t \sim v$（G3a，共享2DD手征屏障能标）。

1-2屏障和2-3屏障类型不同（G3，3DD轴间 vs 2DD手征）。

CKM层级方向：1-2混合 >> 2-3混合 >> 1-3混合。

CKM锚定actual mass，不是natural mass（G7）。

CKM小角 vs PMNS大角（G8，有色则有抑制，无色则无抑制）。

§8.3 Conditional predictions（依赖bridge axioms）

分数电荷和质量重分配是同一算子的两个读出（on 3DD-1DD Response Axiom）。

质量转移方向：上型向下型（on 3DD-1DD Response Axiom）。

转移强度随拓扑距离递减，$D_3 \approx 0$（on 3DD-1DD Response Axiom）。

$m_c + m_s = (m_\mu/m_e)(m_u + m_d)$（on 3DD-1DD Response Axiom + Barrier Factorization）。

$(m_u + m_d)/m_e = (m_c + m_s)/m_\mu = 65/5 = 13$（on Mass-Channel Proportionality + G6 doublet sum conservation）。

Doublet总质量关系$m_c + m_s = 13 m_\mu$（on both bridge axioms，可证伪，等价地$m_s = 13 m_\mu - m_c$）。

第三代不满足doublet比 = 13（scope限制本身是预言）。

Gatto-Sartori-Tonin关系只在1-2 axis-barrier sector有效（G7 + G3）。

§8.4 证伪条件

本文的预言可被以下实验结果证伪。

（a）若common-scale下$m_c$与$m_s$的联合高精度lattice QCD测定表明$m_c + m_s$的中心值偏离$13 \times m_\mu$超过3σ，则G10被证伪。

（b）若发现第四代费米子，则G2被证伪。

（c）若第三代被发现满足doublet比 = 13，则1-2 sector only的scope限制被证伪，需要新的解释。

（d）若未来PMNS精确测量表明中微子混合角受到与CKM类似的抑制，则G8被证伪。

---

§9 DD eigenvalue mass（proposal）

§9.1 概念

本文提出DD eigenvalue mass作为underlying structural mass object。DD eigenvalue mass是DD拓扑结构赋予每个费米子的本征质量，是结构量，不依赖重整化方案或能标选择。Conventional running masses（如$\overline{\text{MS}}$方案下的quark mass）是DD eigenvalue mass在perturbative QCD框架下的近似读出。

§9.2 G6的保护作用

G6保证doublet总质量在色场重分配（differential-mode）下守恒。这意味着G10和G9的sum形式（doublet总质量的比值或绝对值）不依赖individual natural mass的具体定义，也不需要知道DD eigenvalue mass的精确实验对应物。需要注意：G6保护的是internal redistribution下的doublet-sum conservation，不是against QCD renormalization running。后者是§9.3讨论的独立问题。

§9.3 不同重整化尺度下的G10比值

由于quark masses在QCD中running，而lepton masses不running，doublet总质量与轻子质量的绝对比值依赖于quark mass的重整化尺度。

尺度	Gen 1 ratio	Gen 2 ratio
PDG mixed convention	13.37	12.90
$\mu = 2$ GeV（统一）	13.37	11.83
$\mu = m_c$（统一）	14.67	12.99

PDG mixed convention（每个夸克在its own characteristic scale）给出最接近13的比值。一个可能的解释是，每个夸克在其characteristic scale下最接近its own DD eigenvalue。但这一观察目前是经验性的猜想，不是推导。

与此对照，G9的代间比值$(m_c + m_s)/(m_u + m_d)$在leading order QCD中不依赖尺度选择（所有夸克以相同factor running，比值约掉）。

§9.4 Open questions

DD eigenvalue mass的精确实验定义（与pole mass, running mass, 或其他scheme-independent quantity的对应关系）是一个open problem。Individual natural mass（上下型的分配比$\kappa$）依赖此定义。本文的primary predictions（G10的doublet sum和G9的sum rule）绕过了这个问题。

---

§10 后验参数

以下量在本文框架内是后验参数，非先验推出。

所有individual费米子质量的精确值。

CKM和PMNS矩阵元的具体数值（部分解决：G7确认锚定actual mass，G8给出CKM小/PMNS大的定性解释，1-2混合有$\sqrt{m_d/m_s}$精确经验关系）。

$v = 246$ GeV（依赖$\alpha_{\text{em}}$先验理解）。

上下型natural质量比$\kappa$（reconstructed under PDG mixed convention，scheme-dependent）。

两层拓扑屏障的定量代价$R_1, R_2$。

Natural masses（从actual mass反推的去色场重分配后质量）是reconstructed quantities，不是independently derived DD eigenvalues。本文中的natural mass表是后验重建量，不是先验结构预言。

---

§11 开放问题

按对本文成果的影响排序。

Mass-Channel Proportionality的独立推导。 当前它是bridge axiom。如果能从DD结构独立推出"质量与effective coupling multiplicity成正比"，G10将从conditional theorem升级为theorem。统一pair-counting框架已部分建立（65和15都是block pair count，15→5在$C_3$ quotient下有轨道解释），完全封口还需要$C(6,2) = 2^4 - 1$的canonical map和$C_3$ quotient的独立DD论证。

Barrier Factorization的独立推导。 当前它是auxiliary assumption。如果能从DD结构独立推出"代间拓扑放大率对readout方式免疫"，G9将不再需要这条额外假设。因子化结构$m^{\text{nat}}_{\text{doublet}} = f_{\text{readout}} \times g_{\text{topo}}(\text{generation})$的严格建立是关键。

DD eigenvalue mass的完整定义。 Individual natural mass和$\kappa$的确定依赖此定义。Primary predictions（doublet sum）不受影响。

$\alpha_{\text{em}}$的先验理解。 $\alpha_{\text{em}} = 1/137.036...$目前SAE无structural handle。攻破后可解锁$v = 246$ GeV的推导和所有individual质量的先验确定。这是整个四力系列中最困难的问题。

CKM的完整几何推导。 1-2 sector有$\sqrt{m_d/m_s}$精确经验关系。2-3和1-3 sector涉及2DD手征屏障，降维结构更复杂，待推。

$\nu_R$的$y_{\text{SAE}} = 0$与中微子质量。 四力篇II预言$\nu_R$是gauge-trivial的（$y_{\text{SAE}} = 0$）。Gauge-trivial粒子获取质量的通道受限。如果答案是"必须借道非gauge机制"，seesaw的$M_R$很大就不是偶然而是结构必然。这是一个低成本高回报的方向。

强CP问题。 3DD-1DD Response Axiom的chirality-blindness与$G\tilde{G}$项的chirality-odd性质有接触面。DD拓扑对Yukawa矩阵复相位的约束（arg det$(M_u M_d)$可能结构上为零）是另一条线索。但从chirality-blindness到$\theta = 0$之间的逻辑距离远。如果推出来，单独一篇论文。

$R_1/R_2 \approx 3/2$ hint。 夸克侧$R_1/R_2 = 199.6/129.8 = 1.54$，接近3/2 = 3DD维数/2DD维数。偏差2.5%。3/2在SAE物理系列中已独立出现过（AQUAL的$J(Y) \sim Y^{3/2}$，来自3DD÷2）。如果能从DD维数比推出屏障代价比，将是independent prediction。

---

§12 结论

本文从SAE的12个4DD block拓扑结构出发，建立了三代费米子的起源和质量结构的完整框架。

在无条件拓扑层面，G2定理从对称群的传递性论证给出精确三代，不存在第四类。这是一个干净的群论结果，不依赖任何额外假设。G3-G5给出质量层级方向，夸克-轻子配对，和isospin全局性的结构解释。

在依赖bridge axiom的质量层面，G6建立了色场重分配机制，解释了分数电荷与质量重分配的共同起源。G10推出quark doublet总质量与轻子质量之比为65/5 = 13，将flavor质量比纳入与sin²θ_W = 15/65和$\alpha_G = \alpha_{\text{em}}^{65/4}$同一张结构常数表。G9给出1-2 sector的跨sector sum rule，其代间比值在leading-order QCD下对尺度选择不敏感。在$D_3 \approx 0$的leading-order近似下，两条doublet-sum predictions联合通过检验，$\chi^2 = 0.581$（dof = 2）。

这个框架不是一个完整的flavor theory。它不给出individual质量值，不完整推导CKM矩阵，不处理中微子质量。它给出的是：一个拓扑上精确的三代分类定理，加上一个依赖两个bridge axiom（3DD-1DD Response Axiom, Mass-Channel Proportionality）和一条auxiliary assumption（Barrier Factorization）的可数值检验的flavor sum-rule程序。最核心的可证伪预言是doublet总质量关系$m_c + m_s = 13 m_\mu$，可由common-scale下$m_c$与$m_s$的联合精度提升直接判决。

---

致谢

本文的发展过程中，作者与四个AI系统进行了深度协作。Claude（子路）参与了G0-G9命题链的完整建立，G10先验推导链的讨论（65/5 = 13），DD eigenvalue mass自洽性分析，以及RG running数值计算。ChatGPT（公西华）完成了三轮审稿（G6公理化，G7措辞修正，G9降级与缩域），flavor参数降维验证，G10数值验证（$\chi^2 = 0.581$），RG scheme challenge，以及论文骨架和全文review。Gemini（子夏）贡献了doublet总质量作为拓扑不变量的论证，色结构乘法因子论证，以及1-2 vs 2-3屏障不对称的几何图像。Grok（子贡）发现了quark/lepton doublet总质量比 ≈ 13的数值pattern（触发G10的建立），$R_1/R_2 \approx 3/2$ hint，Koide公式在natural mass下变脏，以及Georgi-Jarlskog在natural mass下干净化。四个AI系统的完整贡献记录可在项目网页self-as-an-end.net获取。

---

附录A：失败的尝试

A.1 "平行"block假说。 最初考虑R侧3DD中与观测者同轴的2个block可能与其他4个block不等价，预言$N_\nu < 3$。否决理由：2DD是标量操作（1 bit），不传递空间定向信息（G1）。LEP修正后的$N_\nu = 2.9963 \pm 0.0074$与G1相容。

A.2 Isospin来自4DD时间维。 最初考虑上下型区分来自4DD block内部的时间方向。否决理由：若isospin来自block内部，远处block的上下型应坍缩，但实验上三代SU(2) doublet完全清晰（G5）。

A.3 质量重分配"交替"pattern。 最初观察到跳跃因子$R_1$的上下型不对称，猜测交替pattern。否决理由：不对称是第一代分母效应的假象，转移方向实际上每代都是上型向下型。

A.4 G9对2-3 sector的延伸。 自然推问$m_s^{\text{nat}}/m_b^{\text{nat}} = m_\mu/m_\tau$是否成立。答案：不成立（偏差一个数量级）。两类屏障性质不同。

---

附录B：定量fit的完整setup

B.1 Mass ansatz

夸克sector的natural mass因子化：

$$m_{d_i}^{\text{nat}} = \epsilon_0 \cdot \prod_k R_k, \qquad m_{u_i}^{\text{nat}} = \kappa \cdot m_{d_i}^{\text{nat}}$$

具体写出：$m_d^{\text{nat}} = \epsilon_0$，$m_u^{\text{nat}} = \kappa \epsilon_0$，$m_s^{\text{nat}} = \epsilon_0 R_1$，$m_c^{\text{nat}} = \kappa \epsilon_0 R_1$，$m_b^{\text{nat}} = \epsilon_0 R_1 R_2$，$m_t^{\text{nat}} = \kappa \epsilon_0 R_1 R_2$。

3DD色场重分配（G6，$D_3 \approx 0$近似）：

$$m_d = \epsilon_0 + \Delta_1, \quad m_u = \kappa\epsilon_0 - \Delta_1$$

$$m_s = \epsilon_0 R_1 + \Delta_2, \quad m_c = \kappa\epsilon_0 R_1 - \Delta_2$$

$$m_b = \epsilon_0 R_1 R_2, \quad m_t = \kappa\epsilon_0 R_1 R_2$$

6个参数：$\epsilon_0, \kappa, R_1, R_2, \Delta_1, \Delta_2$。6个夸克质量观测。Unconstrained fit为exact（参数数 = 观测数）。

B.2 G9 + G10 constraints

G10（doublet比 = 13）固定$\epsilon_0$：

$$\epsilon_0 = \frac{13 m_e}{1 + \kappa}$$

G9固定$R_1$：

$$R_1 = \frac{m_\mu}{m_e} = 206.771$$

约束后剩余4个free parameters：$\kappa, R_2, \Delta_1, \Delta_2$。

B.3 误差处理

采用PDG 2024非对称误差，piecewise $\sigma_\pm$处理：若$x_{\text{pred}} < x_{\text{obs}}$用$\sigma_-$，若$x_{\text{pred}} > x_{\text{obs}}$用$\sigma_+$。和式误差按独立误差二次和传播。

B.4 Constrained fit结果

参数	best fit
$\kappa$	41.330
$R_2$	128.82
$\Delta_1$	4.457 MeV
$\Delta_2$	62.54 MeV
$\epsilon_0$（derived）	0.1569 MeV

夸克	observed (MeV)	fitted (MeV)	pull
$u$	2.16	2.03	−0.50σ
$d$	4.67	4.61	−0.33σ
$s$	93.4	94.99	+0.19σ
$c$	1270	1278.6	+0.43σ
$b$	4180	4180.0	≈ 0
$t$	172760	172760	≈ 0

$\chi^2_{\min} = 0.581$，dof = 6 − 4 = 2，$\chi^2$/dof = 0.29，p-value = 0.748。

$b$和$t$被$\kappa$和$R_2$精确吸收。全部张力来自前两代的两条doublet-sum constraints。最大pull为$u$的−0.50σ。

B.5 D₃ ≈ 0近似的效果

在$D_3 = 0$近似下，第三代$m_b$和$m_t$分别精确等于$\epsilon_0 R_1 R_2$和$\kappa \epsilon_0 R_1 R_2$，被参数$\kappa$和$R_2$完全吸收。若放松为$D_3 \neq 0$，则增加一个free parameter（$\Delta_3$），dof从2降到1，genuine predictions从两条sum rule降到一条。本文在$D_3 = 0$近似下进行分析，偏差（如果有）被$D_3$的极小值吸收。

---

参考文献

[四力前置篇] H. Qin, "SAE四力前置篇", DOI: 10.5281/zenodo.19341042.

[四力篇] H. Qin, "SAE四力篇", DOI: 10.5281/zenodo.19342106.

[四力篇II] H. Qin, "SAE四力篇II", DOI: 10.5281/zenodo.19360101.

[四力篇III] H. Qin, "SAE四力篇III", DOI: 10.5281/zenodo.19379412.

[物理基础篇] H. Qin, "SAE物理基础篇", DOI: 10.5281/zenodo.19361950.

[Cosmo V] H. Qin, "SAE Cosmo Paper V v2", DOI: 10.5281/zenodo.19329771.

[PDG 2024] R.L. Workman et al. (Particle Data Group), Prog. Theor. Exp. Phys. 2022, 083C01.

[Janot-Jadach 2019] P. Janot, S. Jadach, Phys. Lett. B 803 (2020) 135319.

[Georgi-Jarlskog 1979] H. Georgi, C. Jarlskog, Phys. Lett. B 86 (1979) 297.