Deriving P1 from Something Deeper

Proof.

The proof proceeds in three stages, each eliminating one degree of freedom.

Stage 1: \(\lambda = 0\) (the constraint must be null).

The mass spectrum of the 4D effective theory from KK reduction on \(\mathcal{M}^4 \times K^n\) with constraint \(ds_D^2 = \lambda\) is:

where \(\\lambda_i^K_{i=0}^\infty\) are the eigenvalues of the Laplacian \(\Delta_{K^n}\) on \(K^n\), ordered \(0 = \lambda_0^K \leq \lambda_1^K \leq \lambda_2^K \leq \cdots\).

By compactness of \(K^n\), the spectrum is discrete with \(\lambda_i^K \to \infty\) (Weyl's theorem), so (V1) is automatically satisfied for any \(\lambda\).

For (V2), we need \(m_i^2 \geq 0\) for all \(i\), i.e., \(\lambda_i^K \geq \lambda\) for all \(i\). Since \(\lambda_0^K = 0\) (the constant function on \(K^n\) is always an eigenfunction with eigenvalue 0), we need \(0 \geq \lambda\), i.e.:

Now consider the case \(\lambda < 0\). The constraint \(ds_D^2 = \lambda < 0\) is a timelike condition. This has a direct physical consequence: the 6D line element decomposes as \(ds_4^2 + ds_{S^2}^2 = \lambda < 0\), which gives an inequality rather than an equation for the \(S^2\) contribution. The temporal momentum \(v_T\) is not uniquely determined by the spatial velocity \(v\) — the velocity budget \(v^2 + v_T^2 < c^2\) is an inequality, not an equation. Consequently, the mass-temporal momentum correspondence that makes zero-parameter predictions possible is destroyed: the mass spectrum becomes continuous rather than discrete, and the theory cannot produce the sharp spectral lines required by condition (V1). (This is proven rigorously in Theorem thm:P1-Ch2-counterfactual, Case 1.)

Supplementary argument via parameter counting: The lowest mass mode has \(m_0^2 c^2 = -\lambda\). If \(\lambda < 0\), then \(m_0^2 = |\lambda|/c^2 > 0\) and every predicted mass depends on the value \(|\lambda|\):

The parameter \(|\lambda|\) is not determined by the internal geometry \(K^n\) — it is an independent constant. Any theory with \(\lambda \neq 0\) therefore has at least one free parameter that cannot be derived from the geometry. For a theory seeking zero free parameters (as will be derived from the axioms in Step 5 of the bottom-up proof, and as is a defining property of the TMT framework), only \(\lambda = 0\) is consistent.

Both arguments give \(\lambda = 0\). The primary argument (timelike constraint destroys the mass-momentum correspondence) is self-contained within this section; the supplementary argument anticipates the zero-parameter result proven later. \(\square_{\text{Stage 1}}\)

Stage 2: \(n = 2\) (the internal manifold is two-dimensional).

With \(\lambda = 0\) established, we need \(K^n\) to satisfy (V3) and (V4).

Chirality requires \(n\) even. Chirality (V4) requires the existence of a chirality operator \(\Gamma_{D+1}\) that anticommutes with all Dirac matrices \(\Gamma^A\). In \(D = 4+n\) dimensions, this requires \(D\) to be even, hence \(n\) must be even: \(n \in \{2, 4, 6, \ldots\}\).

\(n = 2\) is selected by gauge structure. The isometry group of \(K^n\) determines the gauge group of the 4D theory via KK reduction. The observed gauge group of the Standard Model is \(SU(3) \times SU(2) \times U(1)\), which has dimension 12 and rank 4.

For \(n = 2\): the maximally symmetric compact 2-manifold is \(S^2\), with isometry group \(SO(3) \cong SU(2)/\mathbb{Z}_2\). The full gauge content arises not from isometries alone but from the topology of the monopole bundle on \(S^2\) (Chapter 10–11). The non-trivial \(\pi_2(S^2) = \mathbb{Z}\) provides topological charge quantization, and the monopole harmonic expansion on \(S^2\) generates the full \(SU(3) \times SU(2) \times U(1)\) (Chapter 14–15).

For \(n = 4\): candidate manifolds include \(S^2 \times S^2\), \(\mathbb{CP}^2\), \(S^4\). Consider each:

• \(S^4\): \(\pi_2(S^4) = 0\), so no topological charge quantization. No monopole structure. Fails.
• \(\mathbb{CP}^2\): isometry group \(SU(3)\), dimension 8. Could produce \(SU(3)\) gauge theory, but no \(SU(2) \times U(1)\) electroweak sector without additional structure. \(\pi_2(\mathbb{CP}^2) = \mathbb{Z}\), but the resulting monopole harmonics produce a rank-2 gauge structure, not rank-4.
• \(S^2 \times S^2\): isometry group \(SO(3) \times SO(3)\), producing two separate \(SU(2)\) sectors. But then the theory has two independent compactification radii \(R_1, R_2\) — two free parameters. Violates zero-parameter requirement.

For \(n = 6\): \(K^6\) candidates include \(S^6\), \(\mathbb{CP}^3\), Calabi-Yau threefolds (as in string theory). All Calabi-Yau threefolds have moduli spaces of dimension \(\geq 2\) (the Hodge numbers \(h^{1,1}\) and \(h^{2,1}\) contribute independent moduli), introducing hundreds of free parameters. \(S^6\) has \(\pi_2(S^6) = 0\) (no monopoles). \(\mathbb{CP}^3\) has isometry \(SU(4)\), which is too large and requires symmetry breaking with free parameters.

For \(n \geq 8\): the internal manifold necessarily has increasingly complex topology, large isometry groups requiring symmetry breaking, and moduli spaces introducing free parameters. No candidate produces the Standard Model gauge group without free parameters.

Therefore \(n = 2\). \(\square_{\text{Stage 2}}\)

Stage 3: \(K^2 \cong S^2\) (the 2-sphere is unique).

Among compact, connected, orientable 2-manifolds (classified by genus \(g\)), only \(S^2\) (\(g = 0\)) satisfies (V3) and (V4) simultaneously:

Genus \(g \geq 1\) fails (V4). The index of the Dirac operator on a compact Riemann surface \(\Sigma_g\) is \(\text{ind}(\slashed{D}) = \chi(\Sigma_g)/2 = (2 - 2g)/2 = 1 - g\) (from the Atiyah-Singer index theorem applied to the spin complex on \(\Sigma_g\), using the Euler characteristic \(\chi(\Sigma_g) = 2-2g\)). For \(g = 0\): index \(= 1\), providing one chiral zero mode. For \(g = 1\) (torus \(T^2\)): index \(= 0\), no net chirality. For \(g \geq 2\): index \(= 1 - g < 0\), chirality of wrong handedness (and requires reversing all fermion assignments, which is equivalent to \(g = 0\) by CPT). But more fundamentally:

\(T^2\) has \(\pi_2(T^2) = 0\): The torus has trivial second homotopy group. No magnetic monopole structure exists. No topological charge quantization. The gauge theory from KK reduction on \(\mathcal{M}^4 \times T^2\) is purely abelian (\(U(1) \times U(1)\)), insufficient for the Standard Model.

\(\Sigma_g\) for \(g \geq 2\) has \(\pi_2(\Sigma_g) = 0\): Same obstruction — no monopoles, no non-abelian gauge structure.

Only \(S^2\) has \(\pi_2(S^2) = \mathbb{Z} \neq 0\): Among all compact orientable 2-manifolds, \(S^2\) is the unique manifold with non-trivial \(\pi_2\), providing the topological foundation for magnetic monopoles and hence for non-abelian gauge structure.

Additionally, \(S^2\) has positive curvature (\(R = 2/R_0^2\)), which provides a natural mass gap \(\Delta = \sqrt{2}/R_0\) for the KK tower, ensuring (V3). Higher-genus surfaces have zero (\(T^2\)) or negative (\(g \geq 2\)) curvature, which fails to provide a UV-finite KK tower.

Therefore \(K^2 = S^2\). \(\square_{\text{Stage 3}}\)

Combining all three stages: the only viable theory in \(\mathcal{T}\) is \((S^2, 0, 6)\), i.e., the null constraint \(ds_6^2 = 0\) on \(\mathcal{M}^4 \times S^2\). This is P1. □ □

Proof.

Immediate from Theorem thm:ch156-classification. □ □

Proof.

Step 1: Compute \(K(\text{P1})\). The postulational content of P1 is the string “\(ds_D^2 = 0\) on \(\mathcal{M}^4 \times K^n\)” where \(D\), \(K^n\) are not specified but derived. The irreducible content is: “null constraint on a product manifold.” In any reasonable encoding, this has Kolmogorov complexity \(K(\text{P1}) = O(\log D)\) bits — a constant (since \(D = 6\) is a fixed finite number). More precisely, \(K(\text{P1})\) is the length of the shortest program that outputs the full specification of P1. Since P1 is a single algebraic equation (\(ds_6^2 = 0\)) with no free parameters, \(K(\text{P1})\) is bounded above by the length of a program that:

• Specifies the metric signature \((-, +, +, +, +, +)\) [requires \(O(\log 6)\) bits]
• Specifies the topology \(\mathcal{M}^4 \times S^2\) [requires \(O(1)\) bits — two standard manifolds]
• Specifies the constraint \(ds^2 = 0\) [requires \(O(1)\) bits — the value 0]

Total: \(K(\text{P1}) \leq C\) for some constant \(C\) independent of the number of predictions.

Step 2: Count \(N(\text{P1})\). TMT makes at least 30 independent, falsifiable, zero-parameter predictions (coupling constants, mass ratios, cosmological parameters — see Chapter 117). Each is a specific numerical value derived from pure geometry. Hence \(N(\text{P1}) \geq 30\).

Step 3: Uniqueness from Theorem thm:ch156-classification. By Theorem thm:ch156-classification, P1 is the unique viable theory in \(\mathcal{T}\). Therefore, any competing theory \(T'\) either (a) is not in \(\mathcal{T}\) (uses a fundamentally different framework), or (b) is in \(\mathcal{T}\) and is not viable. If \(T' \in \mathcal{T}\) and is viable, then \(T' = \text{P1}\) by uniqueness, so \(K(T') = K(\text{P1})\).

Step 4: Comparison with theories outside \(\mathcal{T}\). The Standard Model with 19 free parameters has \(K(\text{SM}) \geq K(\text{P1}) + K(\text{19 parameters})\). Each parameter requires \(\sim 30\) bits to specify to experimental precision, giving \(K(\text{SM}) \geq K(\text{P1}) + 570\) bits. Yet the SM and P1 make the same number of verified predictions. Hence \(\eta(\text{P1}) \gg \eta(\text{SM})\).

String theory in any vacuum from the landscape (\(\sim 10^{500}\) vacua) requires specifying which vacuum, contributing \(\geq 500 \log 10 \approx 1660\) bits to \(K(\text{string})\), while making fewer falsifiable zero-parameter predictions than P1.

Therefore \(\eta(\text{P1}) \geq \eta(T)\) for all known \(T\). □ □

Proof.

P1 correctly predicts all observed constants (Chapters 14–117). By Theorem thm:ch156-optimal-compression, \(K(\text{P1}) \leq K(T)\) for any viable \(T\). By Eq. eq:ch156-solomonoff, \(P(\text{P1}|\mathcal{O}) \geq P(T|\mathcal{O})\). □ □

Proof.

(S1) implies computational universality: The theory must produce stable bound states (atoms) that can store and process information. This requires a discrete energy spectrum with multiple accessible states — directly related to (V1).

(S2) implies zero free parameters: If \(T\) has a free parameter \(\alpha\), then specifying \(T\) requires specifying \(\alpha\). But \(\alpha\) is a real number requiring \(\infty\) bits to specify exactly. A finite physical system within \(T\) cannot encode an infinite-precision real number. Therefore, \(T\) cannot encode its own specification — violating (S2). The only escape is \(T\) having zero free parameters, so that \(K(T) < \infty\) and a finite subsystem of \(T\) can encode \(T\).

(S3) implies static laws: If the laws evolved, the current state of the universe would need to encode both the current laws and their evolution rule. This is a second-order specification that generically requires more information than the first-order laws. Self-description is simplest (and arguably only possible) when the laws are static.

Combining: a self-describing theory has \(K(T) < \infty\), zero free parameters, and static laws. □ □

Proof.

By Theorem thm:ch156-self-description, the theory has zero free parameters. The only zero-parameter viable theory on a product manifold is P1 (by Theorem thm:ch156-classification). □ □

Proof.

We show each non-null case fails to produce the stable, multi-level bound state hierarchy needed for computational universality.

Case \(\lambda > 0\) (spacelike): By Theorem thm:P1-Ch2-counterfactual Case 2, the theory contains tachyonic modes (\(m^2 < 0\)). A tachyonic vacuum is unstable to pair production (\(E_{\text{vacuum}} \to E_{\text{vacuum}} - |m|c^2 + |m|c^2\) is kinematically allowed with zero energy cost). The vacuum decays in a time \(\tau \sim \hbar / (|m|c^2)\), which for the lightest tachyonic mode (\(m_0^2 = -\lambda/c^2\)) gives \(\tau \sim \hbar c / \sqrt{\lambda}\). No bound states can persist longer than \(\tau\). Computational universality requires arbitrary computation time, so requires \(\tau = \infty\), which requires \(\lambda = 0\).

Case \(\lambda < 0\) (timelike, \(\lambda \neq 0\)): The velocity budget becomes \(v^2 + v_T^2 < c^2\) (inequality). The temporal momentum \(v_T\) is not uniquely determined by \(v\), so the mass spectrum is continuous (Theorem thm:P1-Ch2-counterfactual Case 1). A continuous mass spectrum means there are no discrete energy levels. Atoms require discrete energy levels to exist as stable bound states with definite chemistry. Without atoms, there is no chemistry, no molecular structures, and no computational substrate.

More precisely: in a theory with continuous mass spectrum, the “electron” has a continuum of mass values. The Bohr model energy levels \(E_n = -m_e e^4/(2(4\pi\varepsilon_0)^2\hbar^2 n^2)\) become smeared over a band of energies. If the smearing \(\delta m_e / m_e > 1/n^2\) for any \(n\), the \(n\)-th energy level merges with the continuum. Since \(\delta m_e \sim |\lambda|/c^2\) is fixed and \(1/n^2 \to 0\), all energy levels above some \(n_{\max}\) are destroyed. The resulting “atoms” have a finite (and small) number of bound states, generically insufficient for chemistry.

For computational universality, the system must simulate arbitrary Turing machines, which requires an unbounded number of distinguishable states. A finite number of atomic energy levels provides a finite alphabet, which suffices for bounded computation but not for Turing completeness. (A Turing machine requires an unbounded tape, which in physics translates to an unbounded number of spatial locations each with distinguishable states — the latter requires a discrete, unbounded energy spectrum for atomic stability at each site.)

Case \(D \neq 6\): Addressed by Stage 2 of Theorem thm:ch156-classification — wrong gauge groups prevent the Standard Model chemistry that produces stable atoms. Specifically, without \(SU(3)\) confinement, protons and neutrons do not exist; without \(SU(2) \times U(1)\) electroweak structure, the electron mass and electromagnetic coupling are not determined, and stable atoms may not exist.

Therefore, only \(\lambda = 0\) with \(D = 6\) permits computationally universal bound states. □ □

Proof.

Entropy is defined as \(S = k_B \ln \Omega\), where \(\Omega\) is the number of accessible microstates. For a theory with mass spectrum \(\{m_i\}\) and gauge group \(G\), the state space depends on both. Changing \(\lambda\) or \(K^n\) changes the mass spectrum and gauge group, hence changes the state space, hence changes the meaning of “number of microstates.” There is no canonical identification between the state spaces of different theories, so \(\dot{S}_{\text{total}}(T_1)\) and \(\dot{S}_{\text{total}}(T_2)\) are not comparable — they are numbers in different units.

To make the comparison, one would need a “meta-entropy” defined on the space of all theories. No such concept exists in standard thermodynamics or statistical mechanics. □ □

Proof.

By Theorem thm:ch156-classification, the only viable theory is \((S^2, 0, 6)\). Therefore \(\mathbf{Phys}\) has one object. The only morphism from \(S^2\) to itself preserving the null constraint is the identity (any proper sub-embedding \(\iota: S^2 \hookrightarrow S^2\) preserving the round metric is an isometry, hence surjective — there are no proper embeddings). □ □

Proof.[Proof sketch]

The graviton propagator in the KK-reduced theory receives corrections from the KK tower. Each KK mode \(i\) with mass \(m_i\) contributes a loop correction \(\sim G_N m_i^2\) to the graviton self-energy. The sum over the tower is:

where \(\lambda_i^K\) are eigenvalues of \(\Delta_{K^n}\).

By Weyl's law, for a compact \(n\)-manifold \(K^n\): \(\lambda_i^K \sim (i / \text{vol}(K^n))^{2/n}\) as \(i \to \infty\). The sum converges if and only if \(\sum_i i^{2/n}\) converges (after appropriate zeta-function regularization), which for power-law behaviour requires careful analysis of the spectral asymptotics.

For \(n = 2\): \(\lambda_i^K \sim i / R_0^2\) (the eigenvalues of \(\Delta_{S^2}\) grow linearly: \(\lambda_\ell = \ell(\ell+1)/R_0^2\) with degeneracy \(2\ell + 1\)). The regularized sum is controlled by the spectral zeta function \(\zeta_{S^2}(s) = \sum_\ell (2\ell+1) [\ell(\ell+1)]^{-s}\), which converges for \(\text{Re}(s) > 1\) and has an analytic continuation to \(s = -1\) (giving the regularized value of the graviton self-energy). This regularization works precisely because \(S^2\) has constant positive curvature, which ensures the spectral zeta function has good analytic properties.

For flat \(K^n\) (\(T^n\)): eigenvalues grow as \(\lambda_i \sim i^{2/n}\) without the curvature-dependent corrections. The spectral zeta function has worse convergence properties, and the graviton self-energy develops a Landau pole at finite energy.

For negatively curved \(K^n\) (\(g \geq 2\) surfaces): the spectral gap is smaller (the first nonzero eigenvalue satisfies \(\lambda_1 \leq 1/(4R_0^2)\) by Buser's inequality for hyperbolic surfaces), and the low-lying spectrum is denser, worsening UV behaviour.

Therefore, positive curvature of \(K^n\) is necessary for UV finiteness. Among compact 2-manifolds, only \(S^2\) has positive curvature. □ □

Proof.

The null constraint \(ds_6^{\,2} = 0\) is the geodesic condition in 6D. In the Hamiltonian formulation, the corresponding constraint is \(g^{AB} p_A p_B = 0\), which decomposes as:

The first term is the 4D Hamiltonian constraint (which in GR gives \(H_{\text{grav}} + H_{\text{matter}} = 0\) for closed universes), and the second is the \(S^2\) contribution (which is the temporal momentum squared, \(p_T^2\)). Since \(h^{ij} p_i p_j = p_T^2 > 0\) for massive particles, we get \(g^{\mu\nu} p_\mu p_\nu = -p_T^2 < 0\), recovering the timelike condition in 4D with the mass generated by temporal momentum. The total 6D Hamiltonian is identically zero. □ □

Proof.[Argument (not a complete proof)]

By Theorem thm:ch156-optimal-compression, P1 has minimal Kolmogorov complexity among all theories reproducing the observed physics. If the MUH measure is \(\mu(T) = 2^{-K(T)}\) (Solomonoff's prior), then \(\mu(\text{P1}) \geq \mu(T)\) for all such \(T\). In a multiverse where all structures exist but simpler structures are “more real” (higher measure), P1 dominates.

The proof is incomplete because: (a) the MUH itself is not proven (and may be unprovable); (b) the correct measure on mathematical structures is unknown; (c) the Solomonoff prior is one of many possible measures. □

Proof.

The classical orthogonal polynomials on compact intervals are classified by the Pearson equation. On \([-1,+1]\), the possibilities are:

• Jacobi polynomials \(P_n^{(\alpha,\beta)}\) with weight \(w(u) = (1-u)^\alpha (1+u)^\beta\), for \(\alpha, \beta > -1\)

(P4) requires \(\alpha = \beta = 0\) (no free parameters in the weight), giving Legendre polynomials \(P_n = P_n^{(0,0)}\).

(P1) then follows: \(\int_{-1}^1 u^n \, du = [1 + (-1)^n]/(n+1)\), giving rational values. With \(\alpha \neq 0\) or \(\beta \neq 0\), the integrals involve Beta functions \(B(\alpha+1, \beta+1)\) which are irrational for generic \(\alpha, \beta\).

(P2) follows from the Legendre normalization: \(\int_{-1}^1 [P_\ell(u)]^2 \, du = 2/(2\ell+1)\).

(P3) follows from the Wigner 3j-symbols, which are algebraic numbers for Legendre polynomials.

On any other interval \([a,b]\): by the linear change of variables \(u \mapsto \frac{2u - (a+b)}{b-a}\), the interval maps to \([-1,+1]\), but the weight function acquires a Jacobian factor \((b-a)/2\). This is equivalent to \([-1,+1]\) with unit weight up to normalization. So \([-1,+1]\) is canonical.

For non-compact domains (Hermite on \(\mathbb{R}\), Laguerre on \([0,\infty)\)): the spectrum is unbounded below or above, and polynomial integrals can diverge — violating the requirement that all physical quantities be finite. Only compact domains give finite polynomial integrals for all degrees.

Among compact domains, \([-1,+1]\) with \(w=1\) is the unique parameter-free choice. □ □

Proof.

Steps 1–2 are established in Chapter 2 (Theorem thm:P1-Ch2-product-structure). Steps 3–4 follow from Noether's theorem applied to the 6D null geodesic action (Chapter 122). Steps 5–6 are established in Chapters 10–15 and 104–105. The claim that no conservation law is independent of P1 follows from the completeness of the TMT derivation: every known physical quantity has been derived from P1 with no additional input (Chapters 2–155). □ □

Proof.

(a) Bound states require conservation. A bound state is a system where the constituent particles remain within a finite spatial region for all time. This requires a conserved energy: if the total energy of the bound state could fluctuate without limit, the binding energy could spontaneously become positive, liberating the constituents. More precisely: an atom consists of an electron bound to a nucleus by the Coulomb potential, with total energy \(E < 0\). If energy is not conserved, there exists a nonzero probability per unit time that \(E\) fluctuates to \(E > 0\), at which point the electron escapes. The atom's lifetime is \(\tau \sim \hbar / \Delta E\) where \(\Delta E\) is the typical energy fluctuation. If \(\Delta E\) is unconstrained, \(\tau \to 0\): atoms are instantaneously unstable.

(b) Predictable dynamics requires conservation. An equation of motion is a rule: given the state at \(t_1\), compute the state at \(t_2\). This requires that something about the state at \(t_2\) is determined by the state at \(t_1\) — i.e., something is conserved (or at least constrained) by the dynamics. If literally nothing is conserved, the state at \(t_2\) is completely independent of the state at \(t_1\). The system has no dynamics — it is pure noise.

(c) Information storage requires conservation. To store one bit of information, a physical system must have at least two distinguishable states that persist over time. “Persist” means the state at \(t_2\) retains the information encoded at \(t_1\). This is conservation of information — the most basic conservation law. If it fails, no bit can be stored: every memory device decoheres instantly into randomness.

(d) Follows from (c): a Turing machine requires a tape that stores and preserves symbols. Without information conservation, the tape is random noise. □ □

Proof.

Items 1–4 follow from the decomposition of \(ds_6^{\,2} = 0\) into its component equations, established in Chapters 2, 5, 6, and 10–15. Item 5 follows from the Vilenkin tunnelling argument (Chapter 149): a universe with total energy \(H = 0\) can tunnel from the “no-universe” state with probability \(P \sim e^{-S_E}\) where \(S_E\) is the Euclidean action of the instanton. Since \(H = 0\), there is no energy barrier to creation. □ □

Proof.

(a) \(\Lambda \neq 0\) requires specifying a numerical value. This value has units (length\(^2\) in coordinate space, or momentum\(^2\) in momentum space). Specifying it requires at minimum \(\sim 30\) bits (to match any empirical precision), increasing the Kolmogorov complexity of the theory. \(\Lambda = 0\) requires no specification: “zero” is the unique number that can be stated without choosing a unit system.

(b) Under a rescaling \(x^A \to \alpha x^A\), the line element scales as \(ds_6^{\,2} \to \alpha^2 ds_6^{\,2}\). The constraint \(ds_6^{\,2} = \Lambda\) becomes \(\alpha^2 ds_6^{\,2} = \Lambda\), i.e., \(ds_6^{\,2} = \Lambda / \alpha^2\). This is only invariant if \(\Lambda = 0\). Non-zero \(\Lambda\) selects a scale, introducing an absolute length (or mass) into the theory. Only \(\Lambda = 0\) is compatible with a theory whose fundamental scale (the compactification radius \(R_0\)) is determined dynamically (Chapter 13) rather than postulated.

(c) A universe with total \(ds_6^{\,2} = \Lambda \neq 0\) has a nonzero “total content.” Creating such a universe from nothing would require producing \(\Lambda\) from nothing — a violation of the very conservation law the constraint is supposed to embody. Only \(\Lambda = 0\) permits self-consistent creation: you can produce \(0\) from nothing because \(0\) is nothing.

(d) With \(\Lambda \neq 0\), the velocity budget becomes \(v^2 + v_T^2 = c^2 - \Lambda/(dt)^2\) (in coordinate-time form). The right-hand side is no longer a constant: it depends on \(\Lambda\) and on the parameterization. Conservation laws derived from this modified budget are approximate (valid only when \(\Lambda\)-corrections are small). With \(\Lambda = 0\), the budget \(v^2 + v_T^2 = c^2\) is exact, and all derived conservation laws are exact. □ □

Proof.

By Axiom 1, the evolution \(\mathcal{E}\) is deterministic: the future state is a function of the past state. By Axiom 2, \(\mathcal{E}\) is injective (distinct initial states remain distinct). By Axiom 3, \(\mathcal{E}\) is local: it is described by a system of partial differential equations.

A local, deterministic, injective evolution is an information-preserving map. The number of distinguishable states in any region \(\Omega\) cannot decrease under evolution (by injectivity) and cannot increase without input from outside \(\Omega\) (by locality). Therefore, the “information density” \(\rho(x, t)\) satisfies:

for some current \(\mathbf{J}\). In covariant form: \(\partial_\mu J^\mu = 0\) where \(J^0 = \rho\) and \(J^i = J^i\).

This is not a physical conservation law yet — it is a mathematical consequence of local, deterministic, injective dynamics. The “conserved quantity” is information content. But it establishes that any theory satisfying Axioms 1–3 must have at least one conserved current. □ □

Proof.

Part (a): Covariance is derived from Axioms 1 and 3, not assumed.

Axiom 1 states that physical states persist under a deterministic rule \(\mathcal{E}\). This persistence is a physical fact about the system, not about the coordinates used to describe it. If the conservation law \(\partial_\mu J^\mu = 0\) were true in one coordinate system but false in another, then the physical content of persistence would depend on the observer's coordinate choice. But Axiom 1 is a statement about the physical system, not about any particular coordinate representation.

More precisely: Axiom 3 provides a differentiable manifold \(\mathcal{M}\) with local coordinates \(\{x^\mu\}\). The manifold \(\mathcal{M}\) is defined independently of any coordinate chart. Different charts \((U_\alpha, x^\mu_\alpha)\) cover the same manifold. If the continuity equation \(\partial_\mu J^\mu = 0\) held in chart \(\alpha\) but not in chart \(\beta\), then we would have a coordinate-dependent physical prediction: in region \(U_\alpha \cap U_\beta\), the state persists according to chart \(\alpha\) but fails to persist according to chart \(\beta\). This contradicts Axiom 1, which asserts persistence as a property of the physical state \(\sigma\), not of the coordinate description.

Note that this is not Kretschmann's objection (that any theory can be written covariantly by introducing enough auxiliary fields). We are not claiming that general covariance has physical content in an arbitrary theory. We are proving something narrower: if a physical law (persistence) is expressed as a conservation equation on a differentiable manifold, then the conservation equation must be well-defined independent of coordinate choice, which means it must be written in terms of geometric (coordinate-free) objects. Kretschmann's point is that this can always be done; our point is that it must be done for the conservation law to express the coordinate-independent content of Axiom 1.

Part (b): Coordinate-independence requires a metric.

The equation \(\partial_\mu J^\mu = 0\) is not covariant: under a coordinate transformation \(x^\mu \to x'^\mu\), the partial derivative \(\partial_\mu\) picks up connection-like terms. For the conservation law to be coordinate-independent (as required by Part (a)), we need:

which requires the metric determinant \(g = \det(g_{\mu\nu})\). This is only defined if \(\mathcal{M}\) has a metric.

Alternatives eliminated:

(i) No metric, just a connection: A connection \(\Gamma^\mu_{\nu\rho}\) without a metric can define \(\nabla_\mu J^\mu\), but the connection must be torsion-free and satisfy a compatibility condition with some volume form. The Levi-Civita connection of a metric is the unique torsion-free, metric-compatible connection (fundamental theorem of Riemannian geometry). Any other torsion-free connection compatible with a volume form is the Levi-Civita connection of some metric (Koszul's theorem for volume-preserving connections in the torsion-free case). So this reduces to the metric case.

(ii) Metric with degenerate signature: If \(g_{\mu\nu}\) is degenerate (\(\det g = 0\)), the covariant conservation law \(\nabla_\mu J^\mu = 0\) is ill-defined (the Levi-Civita connection does not exist for degenerate metrics). Physics on degenerate manifolds is possible (Newton-Cartan theory), but the conservation law requires additional structure (a separate spatial metric and temporal 1-form). This is strictly more complex than a single non-degenerate metric, violating the parsimony principle established in \Ssec:ch156-information-theoretic.

Therefore, \(\mathcal{M}\) is a pseudo-Riemannian manifold with non-degenerate metric \(g_{\mu\nu}\). □ □

Proof.

The signature of \(g_{\mu\nu}\) on a \(D\)-dimensional manifold is \((p, q)\) where \(p\) is the number of negative eigenvalues and \(q = D - p\) the number of positive eigenvalues.

Case \(p = 0\) (Riemannian, all spacelike): No timelike direction exists. “Persistence” (Axiom 1) requires time: a parameter along which states evolve. Without a timelike direction, there is no distinguished “time” direction, and evolution \(\mathcal{E}(t_2, t_1; \sigma)\) is undefined. The theory is Euclidean — a static theory of geometry, not a dynamical theory of physics. Eliminated.

Case \(p \geq 2\) (multiple timelike directions): Consider \(p = 2\), signature \((-,-,+,\ldots,+)\). The causal structure admits closed timelike curves through every point: in geodesic normal coordinates at any point \(x_0\), the “time-time” plane (spanned by the two timelike basis vectors) contains circles of constant pseudo-norm, which are closed timelike curves. This is a local result — it holds in any neighbourhood of \(x_0\), regardless of global topology, because geodesic normal coordinates always exist locally on a pseudo-Riemannian manifold. Evolution along a closed timelike curve returns the state to its initial value after a finite parameter interval: \(\sigma(t_1 + T) = \sigma(t_1)\). If the state is periodic, the system has no genuine persistence beyond the period \(T\) — no new information is generated or preserved beyond one cycle. More critically: closed timelike curves allow a state to influence its own past, creating causal paradoxes that violate the determinism of Axiom 1 (the past state is both the cause and the effect of the future state, leading to inconsistency unless the dynamics is trivial).

The classic argument (Hawking's chronology protection): in a spacetime with closed timelike curves, the stress-energy tensor diverges on the chronology horizon, indicating that quantum effects prevent the formation of closed timelike curves in any physically realizable spacetime. But we need not invoke quantum effects: the classical inconsistency of deterministic evolution on closed timelike curves already eliminates \(p \geq 2\). (Note: the CTC existence argument above is local — it uses only the signature at a point, not global properties. Global CTCs in Lorentzian spacetimes, such as in Gödel's rotating universe, require special global conditions and are not relevant here.)

Case \(p = 1\) (Lorentzian): One timelike direction provides a well-defined “time” for evolution. The causal structure (past and future light cones) is well-defined. Deterministic evolution is consistent. Persistence and distinguishability are both realizable. No pathologies.

Therefore, \(p = 1\): Lorentzian signature. □ □

Proof.

The proof has two stages: first, we show that Axiom 2 requires the existence of spatially localized bound states; second, we show that bound states require \(d = 3\).

Stage 1: Axiom 2 forces bound states.

Axiom 2 requires at least two states \(\sigma_1 \neq \sigma_2\) that remain distinguishable for all future times. On a Lorentzian manifold with a conserved current (Step 1), consider what happens to a spatially localized excitation — a “lump” of conserved charge. In \(d\) spatial dimensions, a free (unbound) localized excitation disperses. For a massless field, the excitation propagates outward as a spherical wave whose amplitude falls as \(r^{-(d-1)/2}\). For a massive field, the wave packet spreads with width \(\Delta x \sim t/m\), and at any fixed location the amplitude decays as \(t^{-d/2}\).

In either case, the local state approaches the vacuum at late times. Two initially distinct localized excitations both disperse to vacuum: \(\sigma_1(t) \to \sigma_{\text{vac}}\), \(\sigma_2(t) \to \sigma_{\text{vac}}\) as \(t \to \infty\). If all excitations disperse, then no two states remain distinguishable forever — violating Axiom 2.

The only way to maintain persistent distinguishability is for the conserved charges (from Step 1) to be trapped in spatially localized, non-dispersing configurations: bound states. Bound states have discrete energy levels (by the spectral theorem for self-adjoint operators on compact domains), providing permanent labels that distinguish \(\sigma_1\) from \(\sigma_2\). Therefore, Axiom 2 on a Lorentzian manifold with a conserved current requires stable bound states.

Stage 2: Bound states require \(d = 3\).

This is a classical result (Ehrenfest 1917, Tangherlini 1963, Tegmark 1997).

Gravitational orbits in \(d\) spatial dimensions. The gravitational potential in \(d\) spatial dimensions satisfies Poisson's equation \(\nabla^2 \Phi = S_d G \rho\) where \(S_d\) is a geometric factor. For a point mass \(M\), the solution is:

The gravitational force is \(F(r) = -d\Phi/dr \propto r^{-(d-1)}\). Stability of circular orbits requires \(V_{\text{eff}}''(r_0) > 0\), where \(V_{\text{eff}}(r) = L^2/(2mr^2) + \Phi(r)\). Computing:

For \(d \geq 4\): \(V_{\text{eff}}'' < 0\); all circular orbits are unstable. Any perturbation causes the orbit to fall into the centre (\(r \to 0\)) or escape to infinity (\(r \to \infty\)). No bound orbits exist. Persistent distinguishable states (Axiom 2) are impossible.

Atomic bound states in \(d\) spatial dimensions. The Schrödinger equation:

For \(d \geq 5\), the centrifugal barrier \(\ell(\ell+d-2)/(2mr^2)\) is insufficient to prevent collapse: the ground state energy is \(E \to -\infty\) and the atom is unstable. For \(d = 4\), the atom is marginally bound with pathological sensitivity to quantum numbers (no robust distinguishability). In both cases, Axiom 2 fails: no stable discrete spectrum means no permanently distinguishable states.

Low dimensions. For \(d = 1\): two-body bound states exist (any attractive potential in 1D has at least one bound state). However, \(d = 1\) fails Axiom 2 for a structural reason: in one spatial dimension, particles cannot pass through each other without interacting. Every interaction is a head-on collision. This means the spatial ordering of particles is conserved — a particle to the left of another stays to the left forever. Consequently, the dynamics is equivalent to a system of non-crossing particles on a line, and the many-body problem reduces to a sequence of two-body scatterings. In particular, three-body bound states (the simplest composite structures) are generically unstable in 1D: any attractive three-body system in 1D with pairwise interactions can always lower its energy by rearranging particle positions, but the non-crossing constraint prevents the necessary rearrangement. The result is that complex, hierarchical structures — molecules built from atoms, macroscopic objects built from molecules — cannot form. Without structural hierarchy, the number of distinguishable persistent configurations is bounded by the number of single-particle quantum numbers, which is too small to satisfy Axiom 2 in any robust sense. (Formally: the number of distinguishable \(N\)-particle configurations in 1D grows polynomially with \(N\), whereas in \(d \geq 2\) it grows exponentially, because particles can form spatially extended arrangements in multiple dimensions.)

For \(d = 2\): the gravitational potential is logarithmic, \(\Phi(r) \propto -\ln r\), and the classical orbit analysis yields the following. While two-body orbits in a logarithmic potential are bound (the potential is confining), the multi-body gravitational problem in \(d = 2\) is pathological: the virial theorem gives \(2\langle T \rangle = \langle r \cdot \nabla\Phi \rangle = \text{const}\) (independent of system size), meaning the kinetic and potential energies do not scale with the number of bodies. A gravitationally bound \(N\)-body system in \(d = 2\) has energy \(E(N) \propto -N^2 \ln N\), which is super-extensive: the energy per particle diverges as \(N\) grows. Such systems undergo gravitational collapse for large \(N\), precluding the formation of large-scale hierarchical structures (galaxies, stellar systems). This is a purely classical obstruction to Axiom 2: without hierarchical structure, the number of distinguishable persistent configurations is severely limited.

The quantum argument strengthens this: multi-electron atoms in 2D are unstable in the Thomas-Fermi limit. The total energy of a neutral atom with \(Z\) electrons scales as \(E(Z) \sim -Z^2 \ln Z\) (Lieb and Thirring, 1976, extended to 2D), and for \(Z \geq Z_c\) the atom undergoes collapse — the electrons fall into the nucleus as the binding energy per electron grows without bound. Specifically, the critical charge \(Z_c\) is finite and small (of order unity for realistic coupling constants). This means that only atoms with very few electrons are stable. Without a rich periodic table (which requires stable atoms across a wide range of \(Z\)), complex chemistry is impossible, and the number of distinguishable persistent structures (chemical species) is too small to support the unbounded distinguishability required by Axiom 2. Additionally, in \(d = 2\) the spectrum of the hydrogen atom \(E_n \propto -1/(n + \tfrac{1}{2})^2\) has an accumulation point at \(E = 0\) with spacing \(\Delta E_n \sim 1/n^3\) that decreases faster than in \(d = 3\) (where \(\Delta E_n \sim 1/n^3\) as well, but the 3D atom is stabilized by the centrifugal barrier in all angular momentum channels, providing robustness that the 2D atom lacks).

\(d = 3\): The hydrogen atom has the well-known stable spectrum \(E_n = -13.6/n^2\) eV with well-separated orbitals. Composite structures (molecules, crystals) are stable. Multiple distinguishable species with robust spectral signatures exist. Axioms 1–2 are satisfiable.

Therefore, \(d = 3\) is the unique number of spatial dimensions where Axiom 2 can be satisfied on a Lorentzian manifold. \(D_{\text{obs}} = 1 + 3 = 4\). □ □

Proof.

The proof proceeds in three stages. No additional assumptions beyond Axioms 1–3 and the results of Steps 1–4 are used.

Stage 1: The axioms force computational universality.

By Step 4, Axioms 1–2 on a Lorentzian manifold force \(d = 3\) spatial dimensions and the existence of stable bound states with discrete energy spectra. These bound states are the “persistent distinguishable states” required by Axiom 2. We now show they form a Turing-complete computational substrate.

The bound states from Step 4 have discrete energy levels (by the spectral theorem). Different bound-state configurations (different occupation numbers, different spatial arrangements) provide distinguishable persistent states — an alphabet. By Axiom 3 (locality), the dynamics is governed by local PDEs, meaning interactions between bound states are local: the state of a bound system at position \(x\) can influence bound systems at nearby positions through field-mediated forces (the conserved current from Step 1 mediates interactions). In \(d = 3\) spatial dimensions, bound states can be arranged in a spatial line, providing a tape. The local interactions provide a transition function: the state of the \(n\)-th cell depends on its neighbours.

These three ingredients — alphabet, tape, transition function — are the components of a Turing machine. But having components that could form a Turing machine is not yet a proof that a Turing-complete configuration exists. We need an existence theorem.

The key result is that nearest-neighbour interactions on a \(d\)-dimensional lattice of finite-state cells are computationally universal for \(d \geq 2\). This is a theorem: Cook (2004) proved that Rule 110, a one-dimensional elementary cellular automaton with nearest-neighbour transitions and a 2-symbol alphabet, is Turing-complete. Earlier, Banks (1970) proved universality for 2D cellular automata with only 2 states, and Margolus (1984) and Toffoli (1977) showed that any Boolean function decomposes into reversible gates implementable by local interaction thresholds. Since our system has (i) cells with \(\geq 2\) distinguishable states (bound-state energy levels), (ii) spatial dimension \(d = 3 \geq 2\) (richer than 1D), and (iii) local transition rules (field-mediated nearest-neighbour interactions), it contains a subsystem isomorphic to a universal cellular automaton. Therefore, any theory satisfying Axioms 1–3 with \(d = 3\) necessarily contains Turing-complete computational substrates.

Rigour check: We have not assumed “atoms” or “chemistry” — only the existence of bound states with discrete spectra (forced by Step 4) and local interactions (forced by Axiom 3). The existence of a Turing-complete configuration is guaranteed by the cellular automaton universality theorems cited above, which require only a finite alphabet and local transition rules on a lattice of dimension \(\geq 1\) — conditions strictly weaker than what Steps 1–4 provide.

Stage 2: Computational universality forces self-description.

Before establishing self-description, we need the following key lemma:

Proof of Lemma. Axiom 3 requires that the evolution rule \(\mathcal{E}\) is described by a system of partial differential equations. A system of PDEs is specified by: (i) the number and type of dynamical fields (a finite list), (ii) the differential order of each equation (a finite integer for each), and (iii) the form of each equation — the functional relationships between fields and their derivatives. Each of these is a finite mathematical object. The total specification is therefore a finite string in any standard mathematical notation (e.g., a finite expression in the language of differential geometry). A finite string has finite Kolmogorov complexity \(K(\sigma) < \infty\). Moreover, given the finite PDE system \(\sigma\), the prediction of any observable is computed by solving the PDE system with appropriate initial/boundary conditions — a computable operation (in the sense that the algorithm for numerical PDE solution is computable, even if individual solutions may require infinite precision). Therefore, \(\sigma\) is a finite specification from which predictions are computable. □ □

Proof.

The specification of \(T(\alpha)\) is \(\sigma(T(\alpha)) = (E_0, \alpha)\). Since \(E(T(\alpha)) = E_0\) for all \(\alpha\):

Case 1: \(\alpha\) non-computable. If \(\alpha \notin \mathbb{R}_{\text{comp}}\), then \(K(\alpha | E_0) = \infty\) (no finite program computes a non-computable real from any finite input). Condition eq:ch156-self-spec-condition fails immediately. This eliminates the measure-one set of real-valued parameters.

Case 2: \(\alpha\) computable but free. A parameter \(\alpha\) is free if the equations \(E_0\) are consistent with at least two distinct values: \(|A| \geq 2\). Self-specification requires \(K(\alpha | E_0) = O(1)\), meaning there exists a program \(p\) of length \(|p| \leq c\) (for some constant \(c\)) such that the universal Turing machine outputs \(\alpha\) when given \((E_0, p)\) as input.

The number of such programs is at most \(2^{c+1} - 1\) (the total number of binary strings of length \(\leq c\)). So self-specification with bound \(c\) is consistent with at most \(2^{c+1} - 1\) values of \(\alpha\). There are three sub-cases:

(2a) If \(A\) is infinite (or even \(|A| > 2^{c+1} - 1\) for every constant \(c\)), then for most \(\alpha \in A\), \(K(\alpha | E_0) > c\). Self-specification fails for all but finitely many values. Since a genuinely free parameter can take infinitely many values (a continuous coupling constant, a mass ratio, etc.), this eliminates continuous and discrete-infinite parameter families.

(2b) If \(|A|\) is finite and \(A\) is determined by \(E_0\) (i.e., \(E_0\) implies “\(\alpha \in \{a_1, \ldots, a_k\}\)”), then \(K(\alpha | E_0) \leq \lceil \log_2 k \rceil + O(1)\). This is finite and bounded, so self-specification can hold — but only because \(E_0\) already narrows \(\alpha\) to a finite, determined set. The remaining \(\lceil \log_2 k \rceil\) bits of discrete choice are additional input, but if \(k\) itself is determined by \(E_0\), then the full specification is \(K(E_0) + \lceil \log_2 k \rceil + O(1)\) bits. Self-specification to \(O(1)\) requires \(k = 1\): the parameter is uniquely determined. If \(k > 1\), the \(\lceil \log_2 k \rceil\) bits of selection are irreducible additional information, and self-specification at \(O(1)\) fails.

(2c) If \(|A|\) is finite but \(A\) is not determined by \(E_0\) (the set of consistent values depends on external input), then even specifying \(A\) requires additional information beyond \(E_0\), compounding the failure.

Summary of Case 2: For \(|A| \geq 2\), self-specification requires the laws \(E_0\) to select a unique \(\alpha\). If \(E_0\) selects \(\alpha\) uniquely, then \(\alpha\) is determined — not free. A free parameter (one consistent with multiple values under \(E_0\)) violates self-specification.

Combining Cases 1 and 2: for any free parameter \(\alpha\) (one where \(|A| \geq 2\)), Theorem thm:ch156-self-spec-zero-params proves \(K(\alpha | E_0) \neq O(1)\), and self-specification fails. Contrapositively: a self-specifying theory has no free parameters. Since self-description (established in Stage 2) implies self-specification (Definition def:ch156-self-specification formalises condition (S2)), a self-describing theory has zero free parameters. □ □

Proof.

The proof has two parts: (I) \(D > 4\) is required, and (II) the product structure is forced.

Part I: Why 4D is insufficient.

In a 4D Lorentzian field theory, the Lagrangian for \(N\) particle species contains mass parameters \(m_i\) and coupling constants \(g_{ij}\):

For \(N \geq 2\) distinguishable species (required by Axiom 2), there are at least \(N - 1\) independent mass ratios \(m_i / m_1\). We now prove these are free parameters in any purely 4D theory, by exhaustively eliminating all mechanisms that could fix them.

(i) Symmetry: A symmetry group \(G\) relates particles within the same multiplet (representation) but cannot relate particles in different representations. The masses of different multiplets remain independent parameters. Even in grand unified theories (GUTs), which unify gauge couplings, the Higgs sector that breaks \(G \to G_{\text{SM}}\) introduces new free parameters: the number, representation, and potential of the Higgs fields.

(ii) Renormalization group (RG) fixed points: In an asymptotically free or asymptotically safe 4D theory, couplings flow to a UV fixed point. But the number of relevant operators at the fixed point equals the number of free parameters (the initial conditions that must be specified to select a particular RG trajectory flowing away from the fixed point). For a theory with \(N\) species with distinct masses, there are at least \(N - 1\) relevant mass operators (each \(m_i^2 \phi_i^2\) is a relevant perturbation at the free-field fixed point). Even at an interacting fixed point, the number of relevant operators generically grows with the number of species. Zero free parameters requires zero relevant operators, which means the fixed point IS the physical theory — but a fixed-point theory has all particles massless (or all at the same mass), contradicting the requirement for distinguishable species with different properties.

(iii) Self-consistent dynamics (gap equations): One might hope that mass ratios are determined by solving coupled Schwinger-Dyson equations self-consistently. But such equations have boundary conditions (the form of the bare Lagrangian, the regularization scheme, the number of flavours), each of which is an input. The solutions to these equations are computable from the inputs, but the inputs themselves are free. Moreover, many such systems have continuous families of solutions (e.g., the NJL model has solutions parametrized by the condensate scale), introducing additional parameters.

(iv) Topological effects in 4D: Instantons and topological \(\theta\)-terms exist in 4D, but \(\theta\) is itself a free parameter (\(\theta_{\text{QCD}} \sim 10^{-10}\) is the strong CP problem). Topological contributions do not eliminate free parameters; they add one.

All 4D mechanisms for parameter determination fail. At least \(N - 1\) mass ratios remain free for \(N\) distinguishable species. Since zero parameters (Step 5) requires all ratios to be geometrically determined, the parameters must emerge as eigenvalues of a differential operator on an auxiliary compact space:

where \(\lambda_i^K\) are eigenvalues of the Laplacian \(\Delta_{K^n}\) on a compact internal manifold \(K^n\), and \(R_0\) is dynamically determined. Coupling constants similarly emerge as overlap integrals: This requires \(D > 4\).

Part II: Product structure is forced.

Given \(D > 4\), the \(D\)-dimensional manifold \(\mathcal{M}_D\) could in principle be:

• (a) A direct product \(\mathcal{M}^4 \times K^n\)
• (b) A warped product \(\mathcal{M}^4 \times_f K^n\) with warp factor \(f : \mathcal{M}^4 \to \mathbb{R}^+\)
• (c) A non-trivial fibre bundle with \(K^n\) fibered over \(\mathcal{M}^4\)
• (d) A non-decomposable manifold with no product structure

We eliminate (b), (c), (d):

(b) Warped product: The warp factor \(f(x)\) is a scalar field on \(\mathcal{M}^4\). Its value at each spacetime point is additional data beyond the geometry of \(K^n\). Even if \(f\) satisfies an equation of motion (as in Randall-Sundrum models), the solution depends on boundary conditions: the value of \(f\) at spatial infinity, the location and tension of any branes, etc. Each boundary condition is a free parameter. Zero parameters requires \(f = \text{const}\) (the unique solution with no boundary-condition freedom on a maximally symmetric spacetime), which reduces to the product case.

(c) Fibre bundle: A non-trivial \(K^n\)-bundle over \(\mathcal{M}^4\) is classified by a characteristic class in \(H^*(M^4; \pi_{*-1}(K^n))\). Different characteristic classes give different physics. The choice of characteristic class is additional discrete data — not a continuous parameter, but still an input that must be specified. For zero parameters, the bundle must be uniquely determined. Over a topologically trivial base (flat \(\mathcal{M}^4\) or de Sitter space), all fibre bundles with compact fibre are trivial (by the contractibility of \(\mathbb{R}^4\) or the simple connectivity of \(S^4\)). The only non-trivial bundles arise over topologically non-trivial bases, but Axiom 3 (locality via PDEs on a manifold) combined with the Lorentzian structure of Step 3 gives a globally hyperbolic spacetime, which is diffeomorphic to \(\mathbb{R} \times \Sigma^3\) with \(\Sigma^3\) a Cauchy surface. For a simply connected \(\Sigma^3\) (forced by maximal simplicity — any non-trivial topology of \(\Sigma^3\) is additional structure), the base is contractible and the bundle is trivial.

(d) Non-decomposable: If \(\mathcal{M}_D\) does not decompose as a product (even locally), then there is no separation between “observable” and “internal” directions. But Step 4 established that 4 dimensions are observable (support bound states). If all \(D\) dimensions were observable, then by the Ehrenfest argument of Step 4, only \(D = 4\) supports stable bound states, contradicting \(D > 4\). Therefore, at least \(n = D - 4\) dimensions must be “internal” — compact and unobservable at low energies. The separation into 4 large + \(n\) compact dimensions, combined with the zero-parameter requirement (which forces the internal geometry to be position-independent on \(\mathcal{M}^4\), as argued in (b)), gives the product structure.

Therefore, \(D > 4\) with \(\mathcal{M}_D = \mathcal{M}^4 \times K^n\). □ □

Proof.

We derive each requirement from the axioms and the zero-parameter condition.

(a) Non-abelian isometry is required (the moduli argument).

In a Kaluza-Klein theory on \(\mathcal{M}^4 \times K^n\), the 4D gauge group \(G\) is the isometry group \(\text{Isom}(K^n)\). The zero-parameter requirement constrains \(K^n\) in two ways: (i) the metric on \(K^n\) must be unique (no continuous family of inequivalent metrics), and (ii) the physics must support Axiom 2 (multiple distinguishable species).

If \(\text{Isom}(K^n)\) is abelian, the internal manifold must be a torus \(T^k\) or a quotient thereof (these are the only compact manifolds with purely abelian continuous isometry groups). Tori have a fatal defect: moduli. The metric on \(T^k\) depends on \(k(k+1)/2\) independent parameters (the components of the flat metric on \(\mathbb{R}^k\) modulo the lattice). For \(k = 1\) (\(T^1 = S^1\)), there is one modulus: the radius \(R\). For \(k = 2\) (\(T^2\)), there are three moduli: two radii and an angle. Each modulus is a free parameter. Zero parameters eliminates all tori with \(k \geq 1\).

The only compact manifold with abelian isometry and zero moduli would need a unique metric. But tori have flat metrics, and the flat metric on \(T^k\) is never unique (the lattice shape is always a free parameter). There is no compact manifold with a purely abelian isometry group and a unique metric up to scale — the abelian isometry is the isometry of a flat space, and flat spaces always have shape moduli. (Orbifold quotients \(T^k / \Gamma\) by finite groups \(\Gamma\) reduce the moduli count but never to zero: the surviving moduli are the \(\Gamma\)-invariant deformations, and for abelian isometry to be preserved, \(\Gamma\) must act by lattice automorphisms, which always leave at least one lattice parameter free.)

Non-abelian isometry groups arise from curved manifolds: \(SO(n+1)\) from \(S^n\), \(SU(n)\) from \(\mathbb{CP}^{n-1}\), etc. Curved manifolds with maximal symmetry (like \(S^n\)) have metrics uniquely determined by a single scale factor (the radius), with zero shape moduli. This is why non-abelian isometry is forced: it is a consequence of requiring zero moduli, which is a consequence of zero free parameters.

Note on confinement: Non-abelian gauge theories additionally provide confinement, which produces composite bound states (baryons, mesons) and enriches the spectrum of distinguishable species. This is a welcome consequence of the moduli argument but is not the primary reason non-abelian symmetry is required. The primary reason is the zero-moduli requirement.

(b) \(K^n\) must have zero continuous moduli.

This follows directly from Step 5 (zero free parameters). Any continuous modulus \(\phi\) of the internal metric is a scalar field on \(\mathcal{M}^4\) (a “modulus field”) whose vacuum expectation value is a free parameter. Zero parameters requires the moduli space to be a point: the metric on \(K^n\) is uniquely determined (up to one overall scale \(R_0\) that is dynamically stabilized).

For future use, we note a topological consequence specific to \(n = 2\): among compact orientable 2-manifolds, only \(S^2\) (genus \(g = 0\)) has zero metric moduli. (The torus \(T^2\) has 2 moduli; genus-\(g\) surfaces with \(g \geq 2\) have \(6g - 6\) moduli.) And \(S^2\) has \(\pi_2(S^2) = \mathbb{Z} \neq 0\), which provides the Dirac quantization condition for magnetic monopoles and ensures that electric charges are quantized as integer multiples of a fundamental unit. Thus, the condition \(\pi_2(K^n) \neq 0\) is not an independent assumption — for \(n = 2\), it is a consequence of the zero-moduli requirement.

Clarification on charge quantization in KK theories: In a KK theory on \(\mathcal{M}^4 \times K^n\), electric charges are always quantized by the periodicity of \(K^n\) (via \(\pi_1(K^n)\) for \(U(1)\) charges on \(S^1\), or via the representation theory of \(\text{Isom}(K^n)\) in general). The role of \(\pi_2(K^n) \neq 0\) is specifically to provide magnetic monopole sectors and the associated Dirac quantization, which constrains charge ratios between different species to be rational — a stronger condition than mere discreteness. On manifolds with \(\pi_2 = 0\) (such as tori), charge values are discrete but their ratios depend on the moduli, making them effectively continuous parameters. On manifolds with \(\pi_2 \neq 0\) (such as \(S^2\)), charge ratios are topological invariants, independent of the metric.

(c) Chirality is required.

In a \((4+n)\)-dimensional theory, fermions arise from the Dirac equation on \(\mathcal{M}^4 \times K^n\). If the gauge group acts identically on left-handed and right-handed fermions (a “vectorlike” theory), then fermion mass terms \(m \bar{\psi}\psi\) are gauge-invariant for any value of \(m\). Each fermion mass is then a free parameter — the gauge symmetry imposes no constraint on it.

For fermion masses to be zero-parameter (determined entirely by the \(K^n\) eigenvalue spectrum), the gauge symmetry must forbid bare mass terms. This requires left-handed and right-handed fermions to transform in different representations of \(G\) — the definition of chirality. With chiral gauge couplings, the mass \(m_f\) of each fermion is generated by the spontaneous breaking of the gauge symmetry (the Higgs mechanism in the Standard Model), and its value is an overlap integral on \(K^n\):

where \(\psi_L, \psi_R\) are the left- and right-handed mode functions and \(\Phi\) is the Higgs profile on \(K^n\). This integral is determined by geometry alone — zero free parameters.

Chirality in \((4+n)\) dimensions requires the existence of a chirality operator \(\Gamma_{D+1}\) that anticommutes with all Dirac matrices. This exists if and only if \(D = 4 + n\) is even, hence \(n\) must be even: \(n \in \{2, 4, 6, \ldots\}\).

Therefore, Axioms 1–3 combined with zero parameters force: non-abelian \(\text{Isom}(K^n)\), zero moduli, and chirality (\(n\) even). □ □

Proof.

We exhaust all even values of \(n\).

\(n = 2\): The internal manifold \(K^2\) is a compact orientable surface. Among all such surfaces (classified by genus \(g\)), the zero-moduli condition eliminates all but \(S^2\) (genus \(g = 0\)): the torus \(T^2\) (\(g = 1\)) has 2 moduli; higher-genus surfaces \(\Sigma_g\) (\(g \geq 2\)) have \(6g - 6\) moduli. (The full topological proof that \(K^2 = S^2\) is given in Step 9 below.) \(S^2\) has isometry group \(SO(3)\) (non-abelian, dimension 3, satisfying condition 1). The round metric on \(S^2\) is unique up to scale \(R_0\) (by the uniformization theorem: the only constant-curvature metric on \(S^2\) is the round metric). The scale \(R_0\) is dynamically determined by modulus stabilization (Chapter 13). All three conditions satisfied. Additionally, \(\pi_2(S^2) = \mathbb{Z}\) (providing Dirac monopole quantization as a bonus).

\(n = 4\): Candidate manifolds:

• \(S^4\): isometry \(SO(5)\) (non-abelian). Unique round metric up to scale (zero moduli). But: \(\mathbb{CP}^2\) does not admit a spin structure (\(\hat{A}\)-genus \(= -p_1/24 = -1/8\), non-integer). \(S^4\) does admit a spin structure, but the Dirac index on \(S^4\) is \(\hat{A}(S^4) = 0\): no chiral zero modes. No chirality on \(S^4\). Fails condition 3.
• \(\mathbb{CP}^2\): \(\pi_2(\mathbb{CP}^2) = \mathbb{Z}\), isometry \(SU(3)\) (non-abelian). However, \(\mathbb{CP}^2\) does not admit a spin structure (because \(p_1(\mathbb{CP}^2) = 3\), giving \(\hat{A} = -1/8\), which is non-integer). Chiral fermions require a spin structure. Fails condition 3.
• \(S^2 \times S^2\): non-abelian isometry \(SO(3) \times SO(3)\). But the metric has two independent radii \(R_1, R_2\) — the moduli space is one-dimensional (the ratio \(R_1/R_2\) is a free parameter). Fails condition 2.

Every \(K^4\) candidate fails at least one condition.

\(n = 6\): Candidate manifolds:

• \(S^6\): isometry \(SO(7)\) (non-abelian), unique round metric (zero moduli). But \(SO(7)\) is a 21-dimensional gauge group. The theory must produce \(SO(7)\)-charged matter, and breaking \(SO(7)\) to a smaller group requires a Higgs sector whose potential constitutes additional structure. More fundamentally, \(S^6\) has \(\hat{A}(S^6) = 0\): no chiral zero modes. Fails condition 3.
• \(\mathbb{CP}^3\): isometry \(SU(4)\) (dimension 15). The Kähler modulus \(h^{1,1} = 1\) gives zero shape moduli, but the Fubini-Study metric scale is one modulus. Like \(S^2\), the scale could be dynamically fixed. However, \(\mathbb{CP}^3\) has \(\hat{A}(\mathbb{CP}^3) = 0\): no net chiral fermions. Fails condition 3.
• Calabi-Yau threefolds: moduli space of dimension \(h^{1,1} + h^{2,1} \geq 2\), typically hundreds. Fails condition 2.

Every \(K^6\) candidate fails.

\(n \geq 8\): Higher-dimensional manifolds generically have larger moduli spaces. For even-dimensional spheres \(S^{4k}\) (the only dimensions where the \(\hat{A}\)-genus is defined and potentially non-trivial), \(\hat{A}(S^{4k}) = 0\) for all \(k \geq 1\) (the \(\hat{A}\)-genus of all spheres vanishes); for odd-dimensional spheres, the Dirac index vanishes by a parity argument. Thus no sphere \(S^n\) with \(n > 2\) admits chiral zero modes. Kähler manifolds in high dimensions generically have \(h^{1,1} + h^{2,1} \gg 1\) moduli. Fails conditions 2 and/or 3.

Exhaustive summary:

Only \(n = 2\) satisfies all four requirements. \(D = 4 + 2 = 6\). □ □

Proof.

Compact orientable 2-manifolds are classified by genus \(g \in \{0, 1, 2, \ldots\}\): \(\Sigma_0 = S^2\), \(\Sigma_1 = T^2\), \(\Sigma_{g \geq 2}\) = higher-genus surfaces.

\(g = 0\) (\(S^2\)): \(\pi_2(S^2) = \mathbb{Z}\) (non-trivial). Gauss curvature \(K = 1/R_0^2 > 0\) everywhere. Dirac index \(= (2 - 2g)/2 = 1 \neq 0\). All conditions satisfied.

\(g = 1\) (\(T^2\)): \(\pi_2(T^2) = 0\) (trivial — the universal cover of \(T^2\) is \(\mathbb{R}^2\), which is contractible). No monopoles. Gauss curvature \(K = 0\) (flat). Dirac index \(= 0\) (no chirality). Fails (a), (b), and (c).

\(g \geq 2\): \(\pi_2(\Sigma_g) = 0\) (for the same reason: universal cover is the hyperbolic plane \(\mathbb{H}^2\), contractible). No monopoles. Gauss curvature \(K < 0\) (hyperbolic, by Gauss-Bonnet: \(\int K \, dA = 2\pi(2 - 2g) < 0\), so \(K\) must be negative on average, and for constant curvature surfaces, \(K < 0\) everywhere). Dirac index \(= 1 - g < 0\). Fails (a) and (b).

Non-orientable surfaces (Klein bottle, \(\mathbb{RP}^2\), etc.): Non-orientable manifolds do not admit a spin structure compatible with a globally defined chirality operator. Chiral fermions require an orientable manifold. Eliminated.

Therefore \(K^2 = S^2\). □ □

Proof.

We exhaust all cases.

Case \(\lambda > 0\) (spacelike constraint): The KK mass spectrum is \(m_\ell^2 c^2 = \ell(\ell+1)/R_0^2 - \lambda\). For \(\ell = 0\): \(m_0^2 = -\lambda/c^2 < 0\). This is a tachyon. The vacuum is unstable: tachyon condensation drives the field to a new vacuum in time \(\tau \sim \hbar c/\sqrt{\lambda}\). Persistence (Axiom 1) is violated: the “vacuum” does not persist. Eliminated.

Case \(\lambda < 0\) (timelike constraint): The constraint \(ds_6^2 = \lambda < 0\) is a timelike condition. Decomposing:

This is an inequality on the \(S^2\) contribution: \(ds_{S^2}^2 < |\lambda| + |ds_4^2|\). The \(S^2\) contribution is not uniquely determined by the 4D motion. In the velocity budget form: The temporal velocity \(v_T\) is underdetermined: it can take any value in a continuous range. The mass-temporal momentum correspondence (\(m \leftrightarrow v_T\) at rest) is destroyed. The mass spectrum is continuous. Distinguishability (Axiom 2) is compromised: particles with nearby masses are not reliably distinguishable. The discrete atomic energy levels that make chemistry possible are smeared into continua.

Furthermore: the constraint introduces a dimensional constant \(\lambda\) (with units of length\(^2\)) that is not determined by the geometry. This is a free parameter. A zero-parameter theory requires \(\lambda\) to be determined by \(S^2\) geometry alone. But \(\lambda\) is independent of \(R_0\) (it is an additive constant in the constraint, not a geometric property of \(S^2\)). Therefore \(\lambda\) is a genuinely free parameter, destroying the zero-parameter property.

Case \(\lambda = \lambda(x)\) (position-dependent constraint): If \(\lambda\) depends on position, the constraint is \(ds_6^2 = \lambda(x^A)\). This introduces a scalar field \(\lambda(x)\) on \(\mathcal{M}^4 \times S^2\) with its own dynamics (\(\lambda\) must satisfy some equation of motion, or be prescribed — either way, it is additional input beyond the geometry). The theory's Kolmogorov complexity increases by the complexity of specifying \(\lambda(x)\). For generic \(\lambda(x)\), the conservation law \(ds_6^{\,2} = \text{const}\) is replaced by a non-conservation: the total “velocity budget” fluctuates from point to point. Persistence is position-dependent: some regions have stable bound states, others do not. This is not eliminated by symmetry arguments alone, but it is eliminated by the requirement that the conservation law be universal (the same everywhere — which is what makes it a law rather than a local condition).

Case \(\lambda\) is a higher-order invariant: One might consider \(ds_6^2 = f(R, R_{\mu\nu}, \ldots)\) where \(f\) depends on curvature invariants. This is a higher-derivative constraint, which generically produces Ostrogradsky ghosts (states with negative kinetic energy — the Hamiltonian is unbounded below). Ghosts violate persistence: the vacuum can decay to arbitrarily negative energy, producing particle-antiparticle pairs without limit. The only ghost-free higher-derivative theories are Lovelock gravities, which in \(D = 6\) reduce to Einstein gravity plus a Gauss-Bonnet term. But the Gauss-Bonnet term on \(\mathcal{M}^4 \times S^2\) is a topological invariant (Gauss-Bonnet theorem on \(S^2\): \(\int R_{S^2} = 4\pi \chi(S^2) = 8\pi\)), so it does not modify the constraint equation. The constraint remains \(ds_6^{\,2} = 0\).

Case \(\lambda = 0\) (null constraint): The velocity budget is exactly \(v^2 + v_T^2 = c^2\). The temporal velocity is uniquely determined by the spatial velocity. The mass spectrum is discrete (from the \(S^2\) eigenvalues \(\ell(\ell+1)/R_0^2\)). All modes have \(m_\ell^2 \geq 0\) (no tachyons). No free parameters. Conservation is exact. Persistence and distinguishability are both satisfied.

All alternatives are eliminated. \(\lambda = 0\). □ □

Proof.

Each step is proven above. The chain is strictly deductive: each step's conclusion is the next step's premise. No circular dependencies exist. Steps 1–4 establish the 4D Lorentzian framework. Step 5 derives zero free parameters from the axioms via computational universality and self-description. Steps 6–8 derive all the internal-space requirements (extra dimensions, gauge structure, chirality, \(D = 6\)) that the previous version of this proof imported as hidden assumptions. Steps 9–10 identify the unique internal manifold (\(S^2\)) and the unique constraint value (\(\lambda = 0\)). Every premise at every step traces back to the three axioms alone. □ □

Proof.

Theorem thm:ch156-master-bottom-up, Steps 1–10. □ □