Lach-Space

A QSN is a countable set N with a relation ≺ whose transitive closure is a partial order, such that the interval [n,m] = {k : n ≺ k ≺ m} is finite for all n, m, and the edge set E = {(n,m) : n ≺ m, no k with n ≺ k ≺ m} gives a locally finite graph.

Read it as "a discrete causal set." The finiteness of [n,m] is the crucial clause: it says there are only finitely many events causally between any two, so the order is locally finite and the graph has a chance of having a sensible continuum limit. Without it, a dense order would already be continuous and there would be nothing to emerge.

A function T : E → [0, ∞).

Read it as "the edge length." This is the axiom that later turns out to carry the metric — the monograph's Part VII proves that g is a function of T and not of the graph. Note what L2 does not say: it does not derive T from the graph's combinatorial structure. The tension is a primitive, and that single fact is the difference between "geometry emerges" and "geometry is recovered."

For an edge e = (n,m), with μ_n, μ_m the out-neighbour measures and d_G the graph distance, K(e) = 1 − W₁(μ_n, μ_m) / d_G(n,m), and K(e) ∈ [−2, 1].

Read it as "Ollivier curvature, discretely." The idea: walk one step from n and from m. If the two resulting probability clouds stay close together, the edge is positively curved (think of a sphere, where geodesics converge); if they pull apart, negatively curved (a saddle). The normalisation by d_G(n,m) makes K dimensionless. Everything in Part III is the statement that this quantity, correctly rescaled, is the Ricci curvature of the continuum limit.

d_eff = inf over paths γ of Σ_{e∈γ} T(e)⁻¹.

Read it as "shortest path, with tension as an inverse length." The reciprocal matters: a large tension makes an edge cheap to traverse, so tension behaves like the square of a metric scale. This is the discrete analogue of a Riemannian length functional, and it is the object whose continuum limit the whole Riemannian sector studies.

E[Q] = Σ_{e∈E} T(e)·K(e), with gradient-flow dynamics.

Read it as "the action." Tension times curvature, summed over edges — the discrete seed of the Einstein–Hilbert action ∫R√g. Part III's Theorem R2 is precisely the statement that this sum converges to that integral, and the proof's one delicate step is why the theory needs an isotropy hypothesis at all.

Each edge carries a self-adjoint operator H_e with discrete spectrum, gap Δ(e) > 0, and τ(e) = 2πħ / Δ(e).

Postulate, and flagged This is the only axiom that is an assumption in the strong sense — not derived, not claimed to be derivable. The natural reading is that the spectral gap of an edge operator sets a local clock rate. An earlier attempt to derive τ from modular flow (KMS theory) is permanently withdrawn: it confused a property of a state with a property of a spectrum, and modular flow is trivial in the trace state of a type II₁ factor — which, as Part VI shows, is exactly the algebra this theory produces. The postulate is honest about being a postulate.

Under L1, L2, L4, with 0 < T_min ≤ T(e) ≤ T_max, the identity map is a bi-Lipschitz equivalence between (N, d_G) and (N, d_eff) with constant T_max/T_min.

ProofEach term of the sum lies in [T_max⁻¹, T_min⁻¹]; summing over a path of length len(γ) and taking infima over paths gives the two-sided bound. ∎

Under L1, L2 with T_max/T_min bounded, H1′ (degree bound D and bilateral tension bounds) and H4′ (two-sided isoperimetry with isoperimetric dimension N₀ ≥ 3), there is a constant C_dbl = C_dbl(D, N₀, T_max/T_min) with

ProofRetained from v3 Lemma 5.3, unconditional. The five-step argument uses H4′ and H1′ only; step D of the proof states explicitly that no curvature hypothesis is used. The classical inputs are the Coulhon–Grigor'yan Nash inequality from isoperimetry, the Carlen–Kusuoka–Stroock Nash-to-ultracontractivity theorem, and the Davies–Delmotte Gaussian bounds — each applied under bounded geometry, not curvature. ∎

Correction recordedThe v3 dependency matrix listed this lemma as consuming K ≥ 0. That was wrong, and the error mattered: it made the theory's strongest unconditional result look curvature-dependent and hid the existence of U3. The corrected row is in RECONCILIATION.md.

Let (Q_n) be QSNs satisfying L1–L5, H1′ with fixed D and fixed T_max/T_min, H3 (finite rescaled diameter) and H4′ with fixed constants, and |N_n| → ∞. Then the rescaled metric measure spaces (N_n, ε_n d_G, μ_n) are precompact in the measured Gromov–Hausdorff topology.

Reading itPrecompactness is an existence claim, and only an existence claim. It says a subsequence converges to some compact metric measure space — not that the limit is a manifold, not that it has the right dimension, not that it carries any curvature bound. This is deliberately the weakest possible continuum-limit statement, and stating it this weakly is what makes it unconditional.

ProofU2 gives doubling with a constant independent of n. H3 bounds the rescaled diameter uniformly. Gromov's precompactness theorem — with the measured refinement of Fukaya and Cheeger–Colding — then gives mGH precompactness; because the measures are probability measures on compact spaces, there is no mass escape. The limit measure inherits the same doubling constant. ∎

The honest continuum limitEverything the theory proves about its limit unconditionally is this: a limit exists, and it is doubling. Every further claim — manifold, smooth, Riemannian, Einstein — requires a hypothesis, and Parts III–IV are the accounting of which ones and what they cost.

On a finite QSN with the L5 energy, the Wetterich equation admits an exact solution; no truncation is involved.

Reading itThe renormalisation-group flow is usually truncated to be solvable at all — one throws away all but finitely many couplings and hopes the discarded ones did not matter. On a finite graph the flow closes exactly, because the space of couplings is finite-dimensional to begin with. This is the one place in the renormalisation sector where the theory does not approximate.

Proofv3 Theorem 10.4, retained. ∎

The mGH limit (X, d_X, m) of U3 is a smooth N₀-manifold with a C³ Riemannian metric g, and m is a positive multiple of the Riemannian volume.

ForeshadowingPart VII proves this hypothesis is false as stated: a sequence satisfying every other hypothesis can have a limit with a conical singularity. Read the theorems below with that in mind — they are true, and they are true on the smooth locus. The Riemannian sector's results survive; H10 does not.

Assume H10, and assume the sequence carries a tension tensor isotropic in the sense H12, with the bias hypothesis H13 and the recovery hypothesis H14. Then, along edges e whose endpoints converge to a common point x with unit direction v_e,

ProofContinuum step. Ollivier's pointwise asymptotic expansion on a Riemannian manifold (Proposition 6; Example 7 and §8 of Ricci curvature of Markov chains on metric spaces, 2009): for the normalised volume on B(x, ε) with g ∈ C³, κ(x,y) = (ε²/(2(N+2)))(Ric(v,v) + O(ε) + O(d(x,y))). Hypotheses discharged against H10: Ollivier requires g ∈ C³ (H10 gives it), and ε, δ below the injectivity radius (satisfied for all large n since ε_n → 0 and the manifold is compact). Discrete step. U3's mGH convergence supplies that the discrete neighbourhoods converge to the manifold neighbourhoods, so the expansion applies at each limit point. ∎

Where the citation stops — H13(b)Classical E proves the continuum expansion to O(ε³). It does not prove that the discrete-minus-continuum discrepancy has coefficient exactly β′·Ric(v̂,v̂). Gray's density expansion fixes the order O(ε²) of the deviation, but the coefficient may contain a scalar-curvature term β″R (harmless — a finite renormalisation of G) and, for non-isotropic laws, an anisotropic term (not harmless). So R1 is modulo Classical E for the continuum step and conditional on H13(b) for the discrete step. The tag is the minimum of the two. This stop is one of the three the monograph carries; the others are E-Lor (Part IV) and H-BE (below).

Why this is not a synthetic RCD claimAn earlier build tried to pass an Ollivier bound to the limit as a synthetic RCD condition and correctly proved that it fails. R1 does not attempt that. It uses Ollivier curvature only in its valid role: a pointwise expansion on a smooth target. The hypothesis that makes this work is H10 — smoothness of the limit — not a synthetic curvature condition.

Under H10, H12, H13, H14, the rescaled QSN action converges:

ProofR1 gives the pointwise integrand. The Riemann sum Σ T K converges to ∫ Ric(v,v) √g because the graph is a mesh-ε_n discretisation of the compact manifold, the integrand is continuous, and the measure converges weakly (U3). The identification of ∫Ric(v,v) with ∫R√g uses the classical sphere-average identity: the average of Ric(v,v) over the unit sphere in each tangent space is R/N₀. The normalisation α_d absorbs the 1/N₀. Variation in g gives the Einstein–Hilbert Euler–Lagrange equations by the Palatini identity. ∎

What this replacedAn earlier attempt proved this by Γ-convergence of functionals on varying spaces — a much harder route that would be needed only if the limit were not known in advance. Since H10 assumes the limit is a smooth manifold, the convergence is an ordinary Riemann-sum convergence on a known target. The distinction matters: Γ-convergence is the right tool when you must discover the limit; here the limit is assumed.

Conditional on H10, H12, H13, H14, the Riemannian sector of Lach-Space reproduces general relativity in the continuum limit, with the cosmological constant Λ_b a boundary condition of the flow.

ProofFrom R2 by the Palatini identity (§Pal(iii) of CITATIONS.md). R2 gives S_n → (1/16πG)∫(R−2Λ_b)√g + ∫L_mat√g; the first variation of that functional in g is

What "recovered" means — read Part VII alongside thisR3 says the Riemannian sector reproduces GR given H10, H12, H13, H14. Part VII shows the metric is not emergent (03) and H10 is false as stated (02). So R3's honest form is: given a network whose edge data already encodes a metric, the continuum limit of that network is general relativity. That is a consistency theorem, and a real one — it is not emergence.

The limit satisfies RCD(K, N₀).

Assume H-BE. Then X decomposes as X = R ⊔ S with R open dense and a topological N₀-manifold, S closed with Hausdorff dimension ≤ N₀ − 2, and every point of R has a neighbourhood bi-Hölder homeomorphic to a Euclidean ball.

Reading itThis is Cheeger–Colding structure theory: a synthetic curvature bound alone buys you a topological manifold on a dense open set, with Hölder-continuous (not smooth) charts, and a singular set of codimension at least two. It is a genuinely weaker conclusion than R1–R3's, and it is obtained from a genuinely weaker hypothesis.

Why H-BE cannot be derived — three independent obstructions(a) No theorem converts a graph Ollivier bound to a graph Bakry–Émery estimate; the two notions are incomparable on general graphs. (b) Mosco convergence of Dirichlet forms controls integrated energies, not the carré du champ, so a graph gradient estimate would not pass to the limit. (c) No finite-dimensional CD(K,N) follows from Ollivier curvature, and RCD(K,∞) plus Ahlfors regularity does not imply RCD(K,N₀) — the weighted manifold (M^{N₀}, g, e^{−V}vol) with Ric + Hess V ≥ K and non-constant V is a counterexample. H-BE is therefore assumed, and R4 is strictly weaker in reach than R1–R3: it gives manifold structure on the regular set only, without smoothness.

Let W₁^τ be the Lorentzian causal 1-Wasserstein cost between probability measures supported in a compact causal region, defined as the supremum over causal couplings:

(i) The map (ν, ν′) ↦ W₁^τ(ν, ν′) is convex, being a supremum of affine functions over a convex constraint set.

(ii) Consequently, from a uniform lower bound K_τ ≥ k on the timelike Ollivier curvature, no contraction estimate W₁^τ(νP, ν′P) ≤ (1−k) W₁^τ(ν, ν′) follows.

(iii) Consequently, no synthetic timelike Ricci lower bound in the TCD sense follows from a uniform lower bound on the discrete timelike Ollivier curvature.

Proof(i) For each causal coupling π, the map (ν,ν′) ↦ ∫τ dπ is affine, and the constraint set Π_≼(ν,ν′) is convex under mixtures. A supremum of affine functions over a convex set is convex — the standard variational characterisation. This is precisely the mechanism by which W₁^τ differs from Riemannian W₁, which as an infimum is not convex in the same way. (ii) The Riemannian contraction (Ollivier, Prop. 20) is proved by gluing two couplings along a geodesic, and the gluing step uses the triangle inequality for W₁. W₁^τ has no triangle inequality, being a cost supremum. (iii) Immediate from (ii), since the TCD condition is characterised by contraction of the heat flow in W₁^τ. ∎

Scope, stated honestlyL1 rules out the contraction route from a discrete timelike Ollivier lower bound to a synthetic timelike Ricci bound. It does not rule out every conceivable route, and it does not assert W₁^τ is never a metric — on a globally hyperbolic spacetime with the causal cost it coincides with a metric in special situations. What is proved is that the mechanism the theory used is unavailable, which is what forces the one-sided hypothesis H6⁺ and the Wick-rotation conclusion below.

On a smooth Lorentzian manifold, the shell Ollivier curvature satisfies

ProofRetained from v3 Lemma 7.4, with the four-step derivation in O'Neill conventions and the de Sitter sign check. ∎

The missing half — Conjecture E-Lor, the second citation stopThe equality form K^{(r)} = −c·Ric + O(r³) is Conjecture E-Lor, not a published theorem. It requires an upper bound on the supremum W₁^τ, hence a near-optimal dual pair from Lorentzian Kantorovich duality whose error is smaller than the signal r²ℓ·Ric. The natural choice f = g = τ(x̄, ·) carries a Hessian error of order r²ℓ/ℓ₀ with ℓ₀ ≤ inj, which is not smaller than the signal. By L1 this is not a technical gap: the dual side cannot be supplied by a contraction argument, because no contraction exists in this signature.

Any hypothesis on QSNs intended to encode a timelike Ricci lower bound must be an upper bound on the timelike Ollivier curvature: ε_n⁻² K_τ ≤ K⁺.

ProofFrom L3 by sign reversal. L3 gives K^{(r)} ≤ −c·Ric(e₀,e₀) + O(r³…) with c > 0, so a lower bound Ric(e₀,e₀) ≥ k yields an upper bound K^{(r)} ≤ −c·k + O(r³…). A hypothesis stated as a lower bound on K_τ is therefore implied by nothing in the geometry and is contradicted by any spacetime with positive timelike Ricci curvature; the consistent direction is the upper bound. By L1 the contraction route that would have supplied the opposite direction does not exist in this signature. ∎

WithdrawnThe original hypothesis H6, stated as a lower bound on K_τ, was inconsistent with the geometry and is withdrawn. This is the kind of error that only a sign analysis catches: H6 looked natural by analogy with the Riemannian case, and the analogy is exactly what fails.

Under the bounded-degree hypothesis H1′, the Lorentzian recovery sequence for the Einstein–Hilbert action requires a tension tensor isotropic in the sense M^{μν} ∝ g^{μν}, which is impossible with positive tensions. Consequently the Lorentzian Einstein–Hilbert action is obtained only through Wick rotation, and the physical content at bounded valence is a frame-dependent gravity with parameters (κ_D, a_D, b_D).

ProofThe impossibility of M^{μν} ∝ g^{μν} with positive tensions is Theorem C1 (Part V): a positive combination of causal outer products is positive semidefinite, and no positive semidefinite form is proportional to an indefinite one. The bounded-degree obstruction is v3 Remark 8.16. ∎

The Lorentzian sector of Lach-Space, at bounded degree, is not general relativity. It is a frame-dependent gravity with three phenomenological parameters per degree bound, recovering general relativity only in the double limit of large degree and Wick rotation.

ProofTwo consequences of L4, both immediate. (i) At fixed bounded degree the theory is not GR — the limit functional is frame-dependent, so it is not the diffeomorphism-invariant Einstein–Hilbert functional. This is a difference in kind, not a small correction. (ii) GR is recovered only in the double limit D → ∞ and Wick rotation: D → ∞ removes the frame dependence (L4's obstruction is the finite direction set), and the Wick rotation supplies the indefinite signature that positive tensions cannot. ∎

Let g have Lorentzian signature (1, d−1). Let v_1, …, v_k be future-directed causal vectors at p, and M^{μν} = Σ_i c_i v_i^μ v_i^ν with c_i > 0. Then M is not proportional to g^{μν}.

ProofFor any w, M(w,w) = Σ_i c_i (v_i·w)² ≥ 0, so M is positive semidefinite. Suppose M = λg with λ ≠ 0. For timelike w, g(w,w) < 0, so λg(w,w) < 0 when λ > 0, contradicting M(w,w) ≥ 0; for spacelike w, g(w,w) > 0, so λg(w,w) < 0 when λ < 0, again contradicting. No λ works. ∎

The mechanism is signature, not degreeA positive combination of causal outer products is positive semidefinite; an indefinite form is not. This is unconditional and has nothing to do with bounded degree. It is the sharpest form of the Lorentzian obstruction: you cannot build an indefinite metric out of positive tensions, however clever the construction.

Let the local law of a QSN be a fixed finite stencil S ⊂ ℤ^d \ {0} of edge directions. Then (a) the tension tensor is isotropic only if S spans a round norm, which a finite S does not; (b) hence the mGH limit is a Finsler (normed) space, not Riemannian, and (H14) fails.

Proof(a) The rescaled graph distance of a periodic stencil graph is the word metric with generating set S and edge costs T(s)⁻¹ = θ⁻¹|s|²; its asymptotic cone is the norm with the stated unit ball — proved directly by LP duality in §BB of CITATIONS.md, replacing a book citation that carried an unverifiable theorem number. A polytope with finitely many facets is not an ellipsoid. Bounded jitter moves the costs by a bounded factor, which by continuity of the asymptotic cone cannot turn a polytope into a round ball. (b) A normed space that is not Euclidean is not Riemannian. ∎

The square lattice — the minimal witnessTake (ℤ/nℤ)² with four nearest-neighbour edges and T ≡ 1. Every vertex has degree 4; the tension tensor is Σ_{±e₁,±e₂} v̂⊗v̂ = 2I, exactly isotropic; and the rescaled distance is the ℓ¹ distance, whose unit ball is the square {|x|+|y| ≤ 1}. So H12 holds in its strongest form — equality at every node — and the limit is still not Riemannian. H12 is not sufficient.

The scope error, and the escapeThe original build generalised from this to "bounded degree forbids isotropy." The first clause is C2 and is correct; the final clause is false, and the counterexample is C4 below. The step silently assumed the local law is a fixed finite stencil — which the build's admissible class was, so it never saw the counterexample. It built only inside the class where the obstruction holds. The escape is to leave that class for a random local law.

At a point p, let ν_p be the local law of edge directions. Then the emergent norm at p is the Minkowski functional of K_p = conv(supp ν_p). Consequently:

(a) the emergent metric is Riemannian iff K_p is round for a.e. p — iff ν_p is not supported on finitely many directions;

(b) H12 (second-moment isotropy) is strictly weaker than the needed condition. The correct hypothesis is H12′: ν_p is the uniform measure on the unit sphere S^{d−1} of g at p, for a.e. p.

Proof(a) The norm with unit ball K_p = conv(supp ν_p) is round iff K_p is a ball iff supp ν_p is not confined to finitely many directions. (b) The second moment ∫v̂⊗v̂ dν_p and the convex hull of the support are not related injectively in the needed direction: the square's four directions and the uniform circle have the same second moment 2I but different convex hulls — a square versus a disk. H12′ ⟹ H12, since the uniform measure has second moment (1/d)g⁻¹ (the tight-frame identity). ∎

Why this matters for R1–R3The sphere-averaging step in R2 — the average of Ric(v̂,v̂) over directions equals R/N₀ — is valid exactly when the direction law is uniform. Over a square it gives (Ric(e₁,e₁)+Ric(e₂,e₂))/2, which is not R/N₀ in general. So H12 must be replaced by H12′ in every result that consumes it.

There exists a sequence (Q_n) satisfying L1–L5, H1′, H3, H4′, H12′, H13, whose mGH limit is isometric to (M, g). Hence (H14) holds for (M, g), and O2 has a positive solution in the Riemannian sector.

The construction in one sentenceSprinkle points on (M,g) as a Poisson process, join each point to its D nearest neighbours in d_g, set the tension equal to the geometric edge length, and orient the edges by a Morse function. Nothing is tuned: D is a dimension-only constant, and the tension is the distance itself.

Why it works — the local input, provedH12′ (direction isotropy), proved here. Rescale a neighbourhood of a typical point by ε_n. The D nearest neighbours converge in law to the D nearest neighbours of the origin in a unit-intensity Euclidean Poisson process — because the manifold is locally Euclidean at scale ε_n → 0. That Euclidean process is invariant under O(d), so the joint law of the rescaled neighbour directions is O(d)-invariant. The empirical measure of a single D-tuple is not uniform for finite D — but the direction law ν_p that governs the emergent norm is the union of direction sets over all vertices in an O(1) neighbourhood of p, a union of ≍ (δ/ε_n)^d independent O(d)-invariant D-tuples, whose empirical measure converges to the uniform measure on S^{d−1}. Hence ν_p is uniform and H12′ holds. ∎

The limit, and where H12′ upgrades a norm to a metricBy C3 the emergent norm at p has unit ball conv(supp ν_p) = conv(S^{d−1}), the Euclidean ball, since ν_p is uniform. A Finsler metric with round norm everywhere is Riemannian. And this is exactly where H12′ does its work: a priori the per-step cost constant μ(θ) could depend on direction θ — but μ is a functional of an O(d)-invariant local law, so μ: S^{d−1} → ℝ is O(d)-invariant, and the only such functions on the sphere are constants. So μ(θ) = c and the limit distance is c·d_g — a Riemannian (in fact homothetic) metric, not merely a Finsler one. The isotropy of the local law is precisely what upgrades "a norm" to "a constant times d_g." The old H12 would give an isotropic second moment but a possibly anisotropic norm μ(θ); H12′ forces μ constant.

Under H3′ and the named classical inputs I (Combes–Thomas), J (integral representation of the sign function), K (Neuberger), the overlap operator is γ₅-Hermitian, satisfies the Ginsparg–Wilson relation, is exponentially local with localisation length

and has the correct index. Doubler-freedom holds under the additional hypothesis H16 (generalised norm-resolvent convergence).

Proofv3 Theorem 9.1 and Corollary 9.2, retained. ∎

How Nielsen–Ninomiya is evadedThe theorem forbids a chiral lattice fermion that is simultaneously local, chirally symmetric, and translation-invariant. The graph is not translation-invariant — it is a random geometric graph — so the theorem's hypothesis fails and no doublers are forced. This is not a loophole; it is the reason a discrete substrate can carry chirality at all.

Under the named classical inputs N (Kazhdan–Lusztig) and O (Andersen–Paradowski), the category of tilting modules of U_q(su(3) ⊕ su(2) ⊕ u(1)) at q = e^{iπ/ℓ}, ℓ ≥ 4 divisible, is a UMTC with the stated S-matrix; the abelian factor is at M = 6ℓ + 1 (odd, modular); anomaly cancellation holds exactly, with all coefficients zero.

Reading itA modular tensor category is the algebraic skeleton of a 3-dimensional topological field theory, and its S-matrix encodes braiding. The claim is that a quantum-group deformation of the Standard Model's gauge algebra produces, at the right root of unity, a consistent modular category with the SM's anomaly structure. The M = 6ℓ + 1 condition matters: it forces the abelian factor to be odd, which is what makes it modular rather than merely pre-modular — a genuine constraint, not a convenience.

Proofv3 Theorem 16.1 and Theorem 17.1. Anomaly cancellation is conditional on the GUQG construction hypotheses G1–G5. The numerical content is verified in exact arithmetic. ∎

Modulo the named classical input Reuter, the truncated Wetterich flow has a non-Gaussian fixed point with

Proofv3 Theorems 11.1, 11.2, 12.1, retained. The v2.0 values θ_G ≈ 2.3, θ_Λ ≈ −0.3, the SM-shifted (1.05, 0.21), and the stability count 52 < 100 are withdrawn. Under the Standard-Model-shifted truncation the fixed point moves to (0.8632468, −1.2393818). ∎

The one instrumentThe exact-arithmetic verification of the algebra sector is the only machine-checked content in the theory, and it has already earned its keep: it caught one error — an ℓ = 5 loop counterexample (5 ≠ 2) that refuted a v3.0 claim. It should be extended to the NGFP numerical content and the Wick-rotation constant. A theory with one machine-checked sector and three hand-proved ones is not unusual in physics; a theory whose hand-proofs are labelled with the same discipline as its machine proofs is.

The inductive limit of finite edge-mode algebras is a hyperfinite Type II₁ factor, with K₀ ≅ ℤ[1/n].

Reading itA Type II₁ factor is the algebra of observables of a system with a finite density of states but no smallest nonzero observable — a continuum of states with a well-defined trace. It is the natural algebra of a quantum system with infinitely many degrees of freedom but finite entropy per region. The K₀ group ℤ[1/n] — rationals with denominators powers of n — is the theory's version of a dimension group, and it is what makes the algebra hyperfinite rather than merely a factor.

Proofv3 Chapter 14, Props 14.2, 14.3, 14.9, 14.11, Corollary 14.4c, retained and unconditional modulo Classical T. The CPW crossed-product isomorphism is Conjecture 14.6, and edge-mode completeness is Conjecture 14.7. The Page-curve results are conditional on H9. ∎

Why this matters for L6The withdrawn derivation of proper time used modular flow — and modular flow is trivial in the trace state of a Type II₁ factor, which is exactly the algebra M4 produces. So the withdrawal of the modular-flow derivation is not an admission of a gap in the mathematics; it is the observation that the object the derivation needed (a state with nontrivial modular flow) does not exist in the algebra the theory actually generates. The postulate L6 is honest precisely because this was checked.

Let Q_n be a bounded-degree QSN whose local direction law ν is second-moment isotropic (H12), with lattice step ε. Then the graph Laplacian eigenvalue on a plane wave of momentum p is

ProofOn a lattice of step ε with unit direction law ν, the Laplacian on e^{ip·x} has eigenvalue λ(p) = ⟨cos(εp·ŵ)⟩ − 1. Expanding, ⟨cos(εp·ŵ)⟩ = 1 − (ε²/2)⟨(p·ŵ)²⟩ + (ε⁴/24)⟨(p·ŵ)⁴⟩ − …. H12 fixes ⟨(p·ŵ)²⟩ = |p|²/d; for ν uniform on the sphere the fourth moment is ⟨(p·ŵ)⁴⟩ = 3|p|⁴/(d(d+2)). Substituting gives λ(p) = −(ε|p|)²/(2d) + (ε|p|)⁴/(8d(d+2)) + O((εp)⁶), whose quartic-to-quadratic ratio is 1/(4(d+2)). Check: d = 3 gives sin(εp)/(εp) − 1 = −(εp)²/6 + (εp)⁴/120, ratio 1/20 = 1/(4·5) ✓; d = 4 gives 2J₁(εp)/(εp) − 1 = −(εp)²/8 + (εp)⁴/192, ratio 1/24 = 1/(4·6) ✓. Both verified numerically. ∎

For every momentum accessible to any conceivable experiment, ξ_MDR (εp)² < 10⁻³⁰.

Proofεp = ℓ_P p/ħ. At the LHC, p/ħ = 3.6×10¹⁹ m⁻¹, so εp = 5.7×10⁻¹⁶ and (εp)² = 3.3×10⁻³¹. For a 10²⁰ eV cosmic ray, (εp)² = 6.7×10⁻¹⁷. For the time-of-flight channel the fractional speed difference is (3/2)ξ_MDR(εp)² ≈ 5×10⁻⁴⁰, a delay of order 10⁻²² s against a sensitivity of order 10⁻³ s. Every channel is null by more than twenty orders of magnitude. ∎

The QSN is at the conservative endThis is the standard quantum-gravity phenomenology situation, and this theory sits at its least observable corner: the correction is quadratic (not linear) in εp, so it is suppressed by two powers of the Planck scale rather than one, and it is universal, so it produces no species- or polarisation-dependent signal a differential measurement could isolate.

The coefficient ξ_MDR is not determined by the axioms and hypotheses of the theory. Two regulators satisfying the same hypotheses — in particular both satisfying H12 — give different ξ_MDR.

Proof by counterexampleTake d = 3 and two direction laws, both second-moment isotropic. Sphere: ν uniform on S², giving ⟨(p·ŵ)⁴⟩ = |p|⁴/5 for every direction, so ξ_MDR = 1/20. Octahedron: ν uniform on {±e₁, ±e₂, ±e₃}. The second moment is ⟨(p·ŵ)²⟩ = |p|²/3 = |p|²/d, so H12 holds. The fourth moment is (1/3)Σ_μ p̂_μ⁴|p|⁴, which is direction-dependent: it equals |p|⁴/3 along the axes and |p|⁴/9 along the body diagonal. Hence ξ_MDR ranges from 1/12 to 1/36 — both isotropic in the second moment, both satisfying H12, neither equal to the sphere's 1/20. Since the axioms and hypotheses admit both regulators, the theory assigns no unique value to ξ_MDR. ∎

There exists a sequence of QSNs satisfying L1, L2, H1′ (bounded degree), H3, H4′, and H12 whose mGH limit is a topological 2-manifold but not a smooth Riemannian manifold: a flat cone.

The constructionTake the triangular lattice and remove a 60° wedge, gluing the two boundary rays. The result is a cone whose apex has total angle 5π/3 instead of 2π. A small circle around the apex has circumference (5π/3)r, not 2πr — so no neighbourhood of the apex is a Euclidean disk, and the space is not a Riemannian manifold. It is homeomorphic to ℝ², so it is a topological manifold.

ProofL1, L2, H1′: the triangular lattice has degree 6, the cone point degree 5; both bounded, T ≡ 1. H12: the six triangular directions satisfy Σ v̂⊗v̂ = 3I (a tight frame). Away from the apex the cone is locally isometric to the plane, so H12 holds exactly there. At the apex the five directions give Σ v̂⊗v̂ = 3I − v̂(300°)⊗v̂(300°) = [[2.75, 0.433],[0.433, 2.25]], not proportional to I — the apex is per-node anisotropic. But H12 is an integral statement, and the apex is one point of measure zero, contributing n^{−d}·5·ε_n² → 0. So the integral H12 holds while the per-node law fails at exactly one vertex. H3, H4′: the cone is locally compact with bounded rescaled diameter; away from the apex it is locally Euclidean so the isoperimetric inequality holds with the plane's constant. ∎

Let (X, d) be a limit of a QSN sequence satisfying H12, with singular set S of Hausdorff codimension ≥ 2. Then H12 holds at (X, d) even though (X, d) is not smooth.

ProofH12 is a sum over edges weighted by tension and squared step radius, converging to ∫_U g^{μν} dvol_g. The singular set has Hausdorff codimension ≥ 2, so it has d-dimensional measure zero. Edges meeting S within a ball of radius r scale as r^{d−2} against r^d edges in the ball, so the defect's contribution is a fraction r^{−2} → 0. The tensor converges to the smooth value on every U, so H12 holds. ∎

Corollary 02.CH12 is insensitive to curvature concentrated on codimension-2 sets. The theory's only isotropy hypothesis is blind to exactly the defects that make the limit non-smooth. Closing O1 by strengthening H12 in its own terms is therefore impossible: any hypothesis stated as a condition on H12's own object — a tension-weighted direction tensor converging to an integral — is an integral over the same measure, and blind to the same defects. The fix must be a condition on a different object (a per-node or Hausdorff-measure statement), not a strengthening of H12.

Let (X, d) be the mGH limit of a QSN sequence satisfying L1–L5, H1′, H3, H4′, H12, and suppose (X, d) satisfies RCD(K, N) (the synthetic route's hypothesis H-BE). Then exactly one of the following holds: (a) (X, d) is a smooth Riemannian manifold; or (b) (X, d) is a metric space whose singular set has Hausdorff codimension ≥ 2 and is non-empty — and the bound is sharp, codimension 2 being attained (02.A).

ProofThe singular set of an RCD(K,N) space has Hausdorff codimension ≥ 2 (Cheeger–Colding; the RCD generalisation by Gigli, Ambrosio–Gigli–Savaré). If S is empty, (a). If S is non-empty, its codimension is ≥ 2 by the theorem, and 02.A shows the bound is attained, so it cannot be improved to "empty." ∎

Why this ends O1 rather than opening itO1 asked for a proof that (a) holds. 02.A shows (b) is consistent with every hypothesis the theory has. So O1 as stated is false, not merely open. The correct replacement: prove the limit is smooth, OR characterise the singular set and decide whether the physical theory admits it. The theory currently has no axiom that selects (a).

Two QSN sequences with the same abstract graph — same vertices, same edges — and different tension fields have mGH limits that are not isometric.

Proof by constructionTake Q_n = (ℤ/nℤ)² with four nearest-neighbour edges, ε_n = 1/n, and two tension fields. T ≡ 1: the tension tensor is Σ_{±e₁,±e₂} v̂⊗v̂ = 2I, exactly isotropic; the graph distance is unweighted Manhattan, so the limit is the ℓ¹ plane with unit ball the square diamond. T_h = 1, T_v = 2: the graph distance is |Δx| + 2|Δy|, so the limit is the normed plane with unit ball the rhombus (±1,0), (0,±2). A normed plane is isometric to ℓ¹ iff its unit ball is a Euclidean square. The second ball is a rhombus whose adjacent sides have cos = −3/5 ≠ 0, so it is not a square and no orthogonal map carries one ball to the other. The two limits are not isometric. ∎

Corollary 03.A.1The emergent metric is a function of the tension field, not of the graph's combinatorial structure. Since L2 posits the tension as edge data rather than deriving it, the metric is an input to the theory's geometry, not an output of its graph. This is the exact sense in which the continuum limit is recovered: the construction is handed g, in the tension, and reproduces it.

The mGH limit of a bounded-degree QSN with bounded tension is the normed space whose unit ball is the convex hull of {T_e v̂_e}, and the emergent metric tensor, when the limit is Riemannian, is the second-moment tensor

ProofThe graph distance is the infimum of Σ T(e) over paths (L2, L4). Rescaling by ε_n → 0, the distance converges to the norm with the stated unit ball — the asymptotic cone of a graph with a bounded local law is the normed space generated by that law. The ball's boundary near the origin is governed by the quadratic form Σ T(e)² v̂⊗v̂, and the ball is an ellipsoid, hence the norm Riemannian, exactly when this form is isotropic with a round hull (H12′). All higher moments affect the ball's shape away from the quadratic regime, not the metric at the origin. ∎

Proposition 03.C — curvature needs inhomogeneityIf the local law is the same at every node, the limit is a homogeneous normed space, and in the Riemannian case it is flat. Curvature requires the local law to vary from node to node. A stationary graph can produce a norm — flat geometry — but not a curved geometry. So even granting a graph-intrinsic tension, the theory recovers curvature rather than producing it: the field carrying the curvature is the target metric, disguised as edge data.

Let {û_k}_{k=1}^{D} be a tight frame: D directions equally spaced by 2π/D. Then for every rotation angle θ,

ProofWrite û_k = (cos(θ + 2πk/D), sin(θ + 2πk/D)). The frame is the image under a rotation of the θ = 0 frame, and both sums are rotation-covariant: Σ û⊗û transforms as R(Σ û⊗û)Rᵀ, and R((D/2)I)Rᵀ = (D/2)I; Σ û transforms as R(Σ û), and R·0 = 0. Evaluating at θ = 0 gives the stated identities. Verified numerically at θ = 0, 0.37, 1.9, 2.718: the second moment is 5·I to machine precision and the first moment is ~10⁻¹⁶. ∎

The degree is d+1, not d(d+1)/2The earlier threshold D ≥ d(d+1)/2 = 10 in 4D is the count for tuning the tensions to make a generic set of D directions isotropic — the map T ↦ Σ T_e v̂⊗v̂ has rank min(D, d(d+1)/2). A tight frame is not generic: it is isotropic at uniform tension, so no tuning is needed and the count does not apply. The minimal centered tight frames in ℝ^d are the regular simplex (D = d+1) and the cross-polytope {±e_i} (D = 2d). So in 4D the frame route needs only D = 5 — well inside the bounded-degree regime, and below item 01's threshold.

If the stencil is held fixed — the same directions at every node — then the mGH limit is the normed space whose unit ball is the convex hull of those directions, a regular D-gon, which is not a Euclidean disk for any finite D.

ProofA fixed local law has a fixed asymptotic cone, namely the norm whose unit ball is the convex hull of the stencil directions. D equally spaced directions have a regular D-gon as their hull, a disk only for D = ∞. ∎

Let a bounded-degree QSN carry, at each node, a D-fold tight frame of random orientation θ(x) (iid across nodes), with uniform tension T ≡ 1, placed on (M, g) by the C4 prescription. Then H15(d) holds exactly at every node, H12′ holds, and the mGH limit is (M, g) — Riemannian, round.

Mechanism (exact)By 05.1 each node's second moment is (D/2)I regardless of θ(x), so H15(d) holds. The emergent norm's unit ball is K_p = conv(supp ν_p); ν_p is the uniform measure on S^{d−1} (05.E), so K_p is the Euclidean disk and the norm is round. The fixed stencil fails not because its second moment is anisotropic (it is not) but because a fixed orientation makes conv(supp ν_p) a polytope; randomising the orientation makes supp ν_p fill the sphere. ∎

The angular profile f(φ) = mean d(0,x)/|x| is 10-fold modulated with amplitude ≈ sec(π/10) − 1 = 0.051 for the fixed stencil, and flat for the random-orientation graph. The fixed-stencil amplitude is constant (the D-gon defect); the random-orientation amplitude shrinks toward zero. A control — an iid-uniform orientation and a deterministic rotating stencil both approach the same small amplitude, while a fixed stencil with large D does not — confirmed the residual is finite-size geodesic noise, not a polyhedral floor.

The honest residual05.D rests on a shape theorem of C4.1's type for the frame graph — the same class of result (linear chemical-distance growth for bounded-degree random geometric graphs), applied to a new graph. That class is published (Yao–Chen–Guo 2011, continuum; Cox–Durrett 1981, lattice), so the residual is a formalisation gap, not a missing theorem. Its local input (H12′) is stronger here than in C4 (exact rather than neighbourhood-union limit), and its connectivity is verified. No new kind of gap is introduced; the two sectors share the same cited theorem class.

A discrete causal substrate with bounded degree and two-sided isoperimetry has a doubling measure and a precompact family of continuum limits (U2, U3).

If the limit is a smooth Riemannian manifold and a bounded-degree recovery sequence exists, the discrete Ollivier curvature converges to Ricci, the discrete action converges to Einstein–Hilbert, and general relativity is recovered (R1, R2, R3). The recovery sequence exists (C4); the smoothness of the limit is the surviving condition (H10) — and it can fail (02).

O2 positively: an isotropic random geometric graph at bounded degree satisfies every hypothesis of H14 and yields the Riemannian limit — modulo the published continuum shape theorem (Yao–Chen–Guo 2011), transferred by the proved D-NN sandwich of C4.1b.

O1 negatively: H10 is false as stated — a QSN sequence satisfying every hypothesis can have a limit with a codimension-2 conical singularity. The honest geometric claim: the limit is either a smooth manifold or an RCD space with a codimension-2 singular set, and the theory does not decide which.

Matter positively: per-node isotropy H15(d) is achievable at bounded degree with no fine-tuning, by an orientation-randomised tight frame.

No discrete Ollivier lower bound can produce a synthetic timelike Ricci lower bound, because the causal 1-Wasserstein cost is convex rather than a metric (L1). The Lorentzian action is reachable only by Wick rotation, and the bounded-degree Lorentzian theory is a frame-dependent gravity (L4, L5). Proved.

Chiral fermions, the Standard Model gauge structure, and anomaly cancellation follow modulo named classical inputs, with the algebra sector verified in exact arithmetic (M1, M2). The ℓ = 5 S-matrix and the NGFP fixed-point cubic are machine-checked (certificates/).

Proper time is a Postulate (L6). Not derived, not claimed derivable. Postulate.