Skip to main content

Fundamental Theorems

"Mathematics is the language in which God has written the Universe." — Galileo Galilei

Bridge from the Previous Chapter

In the previous chapter we defined all the key concepts of CC: the Holon, six measures (PP, SvNS_{vN}, Φ\Phi, DdiffD_{\text{diff}}, RR, CC), E-coherence, the stress tensor, the interiority hierarchy, and the sensorimotor functors. Those were the "bricks". Now it is time to build the edifice from them — a system of theorems in which each result follows logically from the previous ones, and together they form a closed deductive chain from the axioms to the deepest conclusions about the nature of life and consciousness.

Chapter Roadmap

In this chapter we:

  1. Prove the existence of dynamics — Theorem 6.1: the evolution equation has a solution (section "Existence Theorems")
  2. Show the necessity of self-reference — Theorems 7.1–7.2: viability requires a self-model φ\varphi, iterations converge to Γ∗\Gamma^* (section "Self-Reference Theorems")
  3. Derive a conditional E-requirement — Theorem 8.1: an explicit stationary purity balance and rate/source bounds (section "The No-Zombie Theorem")
  4. Investigate composition — Theorems 9.1–9.3: fractal closure, scale invariance, and when coupling correlates the parts (section "Composition Theorems")
  5. Derive a unified viability criterion — Theorem 10.1: ∥σsys∥∞<1\|\sigma_{\mathrm{sys}}\|_\infty < 1 (section "Unified Viability Condition")
  6. Describe the sensorimotor cycle — Theorems 11.1–11.4: encoding, action, completeness, hedonics (section "Sensorimotor Encoding")
  7. Examine attractors and structure — T-96, T-98, Fano uniqueness (sections "Attractor Theorems", "Fano Uniqueness")

Why do we need a chapter on theorems? We already know the axioms and definitions. But axioms are the foundation of a building, and definitions are the bricks. Theorems are the building itself: logical chains that connect the foundation to the roof and show that the structure will not collapse.

This chapter tells a story. It begins with the question "does dynamics even exist?" (Theorem 6.1), passes through the discovery that every living system must observe itself (Theorem 7.1), derives the conditional dynamical scope of the No-Zombie proposal (Theorem 8.1) — and ends with the question of when the interaction of parts produces something new — a joint state that the parts do not fix (Theorem 9.3: not for every coupling, but when the coupling has a correlating part).

Each theorem is not an isolated fact, but a link in a single deductive chain. Read in order — and you will see how an entire science of life, consciousness, and self-organisation grows from five axioms.

Formalisation Levels

Each result is marked with one of the statuses (complete system — see Status Registry):

  • [T] Theorem — strictly proved from UHM axioms
  • [C] Conditional — conditional on an explicit assumption
  • [H] Hypothesis — mathematically formulated, requires proof or non-perturbative computation
  • [I] Interpretation — a semantic bridge, formally open
  • [D] Definition by convention — a convention
  • [Pr] Programme — a research direction, open problem
A Note on Notation

In this document:


Existence and preservation​

Theorem 6.1: Well-posed initial values [T under explicit hypotheses]​

Let HH be Hermitian, L\mathcal L a finite-dimensional GKSL generator, B(ρ)∈D7B(\rho)\in\mathcal D_7, and a(ρ)≥0a(\rho)\ge0. Assume a locally Lipschitz extension of

F(ρ)=−i[H,ρ]+L(ρ)+a(ρ)(B(ρ)−ρ)F(\rho)=-i[H,\rho]+\mathcal L(\rho)+a(\rho)(B(\rho)-\rho)

to a neighbourhood in Hermitian trace-one space. The initial-value problem has a unique local solution. If D7\mathcal D_7 is invariant, it continues for every finite positive time because the state space is compact and the vector field is bounded there. This theorem concerns the continuous model; it does not guarantee absence of NaN or accuracy of an arbitrary discretisation. Continuity of coefficients alone does not imply uniqueness.

Theorem 6.2: Preservation of density states [T]​

Under the hypotheses above, Hermiticity and trace are preserved. For v∈ker⁡ρv\in\ker\rho, v∗(−i[H,ρ])v=0v^*(-i[H,\rho])v=0, v∗L(ρ)v≥0v^*\mathcal L(\rho)v\ge0 from the CP part of GKSL, and v∗a(B−ρ)v=av∗Bv≥0v^*a(B-\rho)v=a v^*Bv\ge0. The vector field therefore lies in the tangent cone of the positive cone at its boundary. Under the stated regularity the invariance criterion preserves D7\mathcal D_7 and the continuation argument applies. A nonlinear density-preserving law need not be one linear completely positive map.

Self-observation: limits of inference​

Theorem 7.1. Limits of the purity criterion​

P>2/7P>2/7 does not imply an internal informational observation. A state sustained by a fixed input can remain stationary in the selected domain without state-dependent feedback. The formal condition ∥ρ−φ(ρ)∥<ε\|\rho-\varphi(\rho)\|<\varepsilon is always satisfied by identity; it does not prove that the system measures or knows itself. A substantive claim requires an observation law, a certificate available to a register, and a specified control class.

Theorem 7.2: Fixed point of a chosen self-model [T]​

For the selected family Mcoh(ρ)=(1−R(ρ))P(ρ)+R(ρ)I/7M_{\rm coh}(\rho)=(1-R(\rho))\mathcal P(\rho)+R(\rho)I/7, with R=1/(7P)R=1/(7P) and P\mathcal P fixing I/7I/7 and nonexpansive in HS norm,

∥Mcoh(ρ)−I/7∥F≤67∥ρ−I/7∥F.\|M_{\rm coh}(\rho)-I/7\|_F\le\tfrac67\|\rho-I/7\|_F.

Subtract I/7I/7 and use 0≤1−R≤6/70\le1-R\le6/7. Iteration gives geometric convergence and the unique fixed point I/7I/7. This is a distance estimate to the indicated state, not a proof of global contraction between arbitrary inputs. It is not implied by another generator's spectral gap and does not apply to every self-model. The logical support reflector is a separate construction.


The No-Zombie Theorem: conditional dynamical content​

The question is whether a declared dissipative system needs E-dependent regeneration to sustain a declared viability class. This requires rate and source assumptions; it does not follow merely from a nonzero dissipator. The identification of E with interiority is [P/I]. A mathematical dependence on E does not, by itself, refute philosophical zombies or epiphenomenalism.

Theorem 8.1: Purity balance and a conditional E-floor [T under stated bounds]​

Canonical CohE\mathrm{Coh}_E ranges from zero to one. At I/7I/7 it equals 1/71/7. A positive lower bound for a stationary state requires restrictions on the rates and source terms in the balance equation below.

Consider a specified state-valued target τ(Γ)\tau(\Gamma) and dynamics

Γ˙=−i[H,Γ]+γ(PFanoΓ−Γ)+κ(Γ)gV(P)(τ(Γ)−Γ)+Jext(Γ).\dot\Gamma=-i[H,\Gamma]+\gamma(\mathcal P_{\mathrm{Fano}}\Gamma-\Gamma) +\kappa(\Gamma)g_V(P)(\tau(\Gamma)-\Gamma)+\mathcal J_{\mathrm{ext}}(\Gamma).

Write W=Pcoh=∑i≠j∣γij∣2W=P_{\mathrm{coh}}=\sum_{i\ne j}|\gamma_{ij}|^2, h=Tr⁡(Γτ)−Ph=\operatorname{Tr}(\Gamma\tau)-P, JP=2Tr⁡(ΓJext)J_P=2\operatorname{Tr}(\Gamma\mathcal J_{\mathrm{ext}}), and κ=κb+κ0CohE\kappa=\kappa_b+\kappa_0\mathrm{Coh}_E. Then the exact purity balance is

P˙=−4γ3W+2(κb+κ0CohE)gVh+JP.\dot P=-\frac{4\gamma}{3}W+2(\kappa_b+\kappa_0\mathrm{Coh}_E)g_Vh+J_P.

Proof. P˙=2Tr⁡(ΓΓ˙)\dot P=2\operatorname{Tr}(\Gamma\dot\Gamma); the commutator has zero trace contribution, and the Fano channel keeps the diagonal while multiplying off-diagonals by 1/31/3. The remaining terms follow by substitution. No phenomenological identification is used.

Stationary pointwise bound [T]. If W>0W>0, JP=0J_P=0, gV>0g_V>0, h>0h>0 and κ0>0\kappa_0>0 at a stationary state, then

CohE=1κ0(2γW3gVh−κb).\mathrm{Coh}_E=\frac1{\kappa_0}\left(\frac{2\gamma W}{3g_Vh}-\kappa_b\right).

A necessary nonnegative lower bound uses the positive part of this expression. The quantity is state-dependent and can be below 1/71/7; no max⁡{1/7,…}\max\{1/7,\ldots\} may be inserted without another assumption. If κ0=0\kappa_0=0, the equation instead constrains the bootstrap rate and yields no E-bound. Replacing PP by PcritP_{\mathrm{crit}} or dropping the gate changes the equation and needs an explicit inequality direction and hypothesis.

Uniform exclusion theorem [T under explicit bounds]. On a proposed stationary viability class suppose

γ≥γmin⁡>0,W≥Wmin⁡>0,JP≤Jmax⁡,0≤gV≤gmax⁡,h≤hmax⁡,0≤κb≤Kb,0≤κ0≤K0,\gamma\ge\gamma_{\min}>0,\quad W\ge W_{\min}>0,\quad J_P\le J_{\max},\quad 0\le g_V\le g_{\max},\quad h\le h_{\max},\quad 0\le\kappa_b\le K_b,\quad0\le\kappa_0\le K_0,

with gmax⁡,hmax⁡,K0>0g_{\max},h_{\max},K_0>0. Every stationary state in that class obeys

CohE≥cmin⁡:=max⁡{0,4γmin⁡Wmin⁡/3−Jmax⁡2K0gmax⁡hmax⁡−KbK0}.\mathrm{Coh}_E\ge c_{\min}:=\max\left\{0,\frac{4\gamma_{\min}W_{\min}/3-J_{\max}}{2K_0g_{\max}h_{\max}}-\frac{K_b}{K_0}\right\}.

Proof. At stationarity, dissipation is at least 4γmin⁡Wmin⁡/34\gamma_{\min}W_{\min}/3, whereas the positive sources are at most 2(Kb+K0CohE)gmax⁡hmax⁡+Jmax⁡2(K_b+K_0\mathrm{Coh}_E)g_{\max}h_{\max}+J_{\max}. Rearrange. If cmin⁡>1/7c_{\min}>1/7, the proposed class contains no stationary state with CohE≤1/7\mathrm{Coh}_E\le1/7. If cmin⁡>1c_{\min}>1, the class has no stationary state at all. The bounds, including a positive required off-diagonal weight, are substantive premises; DΩ≠0\mathcal D_\Omega\ne0 alone supplies none of them.

Canonical self-model: why preserving some coherence is insufficient​

For τ=φcohΓ=(1−R)DαΓ+RI/7\tau=\varphi_{\mathrm{coh}}\Gamma=(1-R)\mathcal D_\alpha\Gamma+RI/7,

h=−R(P−1/7)−(1−R)αW≤0.h=-R(P-1/7)-(1-R)\alpha W\le0.

Thus this model cannot balance positive Fano purity loss in the source-free equation. Choosing a nonunital anchor, sharpening map, Hamiltonian-driven environment or external injection changes the gain term and must be stated. A failure of a diagonal replacement channel does not prove the uniqueness or necessity of this particular canonical model.

For a fixed target and fixed effective rate k=κgVk=\kappa g_V, a diagonal Hamiltonian gives stationary pair magnitude

∣γij∗∣2=k2∣τij∣2(d+k)2+Δωij2,d=2γ/3,|\gamma_{ij}^*|^2=\frac{k^2|\tau_{ij}|^2}{(d+k)^2+\Delta\omega_{ij}^2},\qquad d=2\gamma/3,

and its derivative with respect to kk is

2k∣τij∣2[d(d+k)+Δωij2][(d+k)2+Δωij2]2≥0.\frac{2k|\tau_{ij}|^2[d(d+k)+\Delta\omega_{ij}^2]}{[(d+k)^2+\Delta\omega_{ij}^2]^2}\ge0.

This monotonicity holds with the indicated quantities frozen. It is not a universal derivative of an endogenous attractor with respect to its own E-statistic. For a general smooth equilibrium branch, sensitivity is determined by the full Jacobian and can change sign or fail at a bifurcation.

Counterexamples to a universal floor​

A nonzero dephasing generator can vanish on a particular diagonal pure state with CohE=0\mathrm{Coh}_E=0. That state remains pure without any regeneration. This refutes a claim based only on P>2/7P>2/7 and D≠0\mathcal D\ne0; it is not asserted to pass an independently required integration gate.

A stronger embodied counterexample permits nonzero coherence on every pair. Choose a full-rank state ρ\rho of purity P>2/7P>2/7 with CohE<1/7\mathrm{Coh}_E<1/7, and let F(ρ)F(\rho) be its Hermitian trace-zero velocity under the specified internal dynamics. For

μ>∥F(ρ)∥op/λmin⁡(ρ),σ:=ρ−F(ρ)/μ,\mu>\|F(\rho)\|_{\mathrm{op}}/\lambda_{\min}(\rho),\qquad\sigma:=\rho-F(\rho)/\mu,

σ\sigma is a density matrix, and adding the legitimate backbone μ(σ−Γ)\mu(\sigma-\Gamma) makes ρ\rho an exact equilibrium. This is a constructive counterexample to excluding such states for all environmental anchors. It does not guarantee stability; a stability claim requires the full Jacobian.

For example let pE=0.01p_E=0.01, other populations (1−pE)/6(1-p_E)/6, vi=piv_i=\sqrt{p_i}, and

ρ=(1−r)diag⁡(p)+rvv†,r2=0.35−∑ipi21−∑ipi2.\rho=(1-r)\operatorname{diag}(p)+rvv^\dagger,\quad r^2=\frac{0.35-\sum_ip_i^2}{1-\sum_ip_i^2}.

It is full rank, every pair is coherently populated, P=0.35P=0.35, Φ>1\Phi>1, and CohE<1/7\mathrm{Coh}_E<1/7. The backbone construction applies. It fails the stipulated proxy D7D≥2D^{7D}\ge2: that gate itself entails CohE≥1/6\mathrm{Coh}_E\ge1/6 by definition. Such definitional exclusion must not be presented as a dynamical no-zombie theorem or as a literal extension-entropy calculation.

Minimal model for a declared balance test​

A reproducible model [D] specifies the Hamiltonian, Fano dissipator, target map, bootstrap/E-dependent rate, gate and independent environmental sources. The source-free special case is obtained by Jext=0\mathcal J_{\mathrm{ext}}=0; the canonical I/7I/7 anchor and a nonunital source are distinct models. The uniform bound requires all of its listed inequalities to hold on the tested class.

Controlled simulation protocol​

  1. Fix the target, environment, rates, gate, E-projection and definition of sustained viability before sampling states.
  2. Verify the exact purity balance and the hypotheses of the proposed uniform bound; distinguish stationary states from finite-time survival.
  3. Compare E-dependent regeneration with bootstrap-only and independent-source controls. E-ablation must preserve positivity and must account for diagonal E-population: removing off-diagonal E entries does not generally make canonical CohE\mathrm{Coh}_E zero.
  4. Report basin dependence, eigenvalues of the tangent Jacobian and uncertainty of parameter bounds. Do not require all random states to share an attractor in a bistable gated system.
  5. A counterexample satisfying all assumptions refutes a mathematical bound. Simulations validate a chosen implementation; biological/phenomenal necessity requires independent measurements and the declared bridge.

No universal 1/41/4 critical exponent or macroscopic rate hierarchy is assumed. At P≤2/7P\le2/7 a source-free nonnegative-gate model with Fano dissipation has P˙≤0\dot P\le0 when its gate is zero; independent injection changes this conclusion.

Corollary 8.1.1: explicit causal dependence [T / I]​

For a declared rate law with κ0>0\kappa_0>0, holding other inputs fixed gives ∂κ/∂CohE=κ0\partial\kappa/\partial\mathrm{Coh}_E=\kappa_0. At points where the target difference and gate are nonzero, changing this input changes the generator. The phenomenological identification of this causal input with experience is [P/I]. Neither strict monotonicity of every stationary purity nor the refutation of every form of epiphenomenalism follows automatically.

Corollary 8.1.2: the scope of “No-Zombie” [C / I]​

Under the uniform bound with cmin⁡>1/7c_{\min}>1/7, the declared stationary class excludes CohE≤1/7\mathrm{Coh}_E\le1/7 [T at the bounds]. Interpreting those excluded states as philosophical zombies additionally requires an E-to-phenomenality bridge [P/I] and a task-specific equivalence of functional behaviour. There is no general theorem excluding all functionally capable systems without phenomenal experience.

Corollary 8.1.3: quantitative E-requirement [T under bounds]​

The explicit cmin⁡c_{\min} above is a model-dependent stationary resource requirement. It is not a universal minimum of experience or a clinical diagnostic threshold. With stronger bootstrap or independent sources it can vanish; with larger required coherence dissipation it can increase. Its empirical interpretation must be calibrated separately.

Composition Theorems​

Let us return to the orchestra analogy. Until now we have been studying one musician (a single holon). Now imagine two orchestras deciding to play together. The first question: will the joint performance be meaningful? The second: will it produce something that was absent from either orchestra individually?

Theorems 9.1–9.6 distinguish closure of a chosen aggregate, retention of its margins at weak coupling, and correlation of the joint steady state. Each has different hypotheses.

Theorem 9.1: conditional persistence under weak coupling​

Assume fixed linear local GKSL generators with unique stationary states ρi0\rho_i^0, and invertible restrictions Li∣Tr=0\mathcal L_i|_{\mathrm{Tr}=0}. For a stationary joint state of L1⊗id+id⊗L2+gLint\mathcal L_1\otimes\mathrm{id}+\mathrm{id}\otimes\mathcal L_2+g\mathcal L_{\mathrm{int}}, partial trace gives

Li(ρig−ρi0)=−gTr⁡jLint(ρ12g).\mathcal L_i(\rho_i^g-\rho_i^0)=-g\operatorname{Tr}_j\mathcal L_{\mathrm{int}}(\rho_{12}^g).

Thus ∥ρig−ρi0∥F≤∣g∣∥Li−1∥∥Tr⁡jLint∥F→F\|\rho_i^g-\rho_i^0\|_F\le |g|\|\mathcal L_i^{-1}\|\|\operatorname{Tr}_j\mathcal L_{\mathrm{int}}\|_{F\to F}, using ∥ρ12g∥F≤1\|\rho_{12}^g\|_F\le1. If twice this bound is smaller than P(ρi0)−2/7P(\rho_i^0)-2/7, both marginals retain structural majority [T at these assumptions]. The actual joint dimension is49. A seven-dimensional aggregation must be declared separately.

The mean marginal aggregate below gives a concrete aggregation map. Positive trace preservation alone supplies no bound on its distance from a chosen reference state.

Theorem 9.2 / T-72 (Scale Invariance, CC-6) [T at weak coupling]​

Conditions for aggregation

The completely depolarizing channel sends every input to I/7I/7. It contracts distances while changing the purity of a pure input from one to 1/71/7. The theorem below therefore uses an explicit comparison bound (AGG). Theorem 9.5 proves such a bound for the specified mean of marginal states at weak coupling. The frame enters Φ\Phi and the Gap profile. A rotation can preserve purity and change both readouts. Strong-coupling counterexample

In Plain Terms

Recall a Russian nesting doll (matryoshka): the small doll resembles the large one, and that resembles an even larger one. Theorem 9.2 says when a holon made of holons keeps the structural properties of its parts (purity, reflection, integration): when the parts are only weakly coupled, and the aggregation, applied to uncoupled parts, returns the state of a part. Then every key invariant of the whole lies within an explicit distance, proportional to the coupling, of the same invariant of a part. With strong coupling nothing of the kind holds: two maximally entangled holons, aggregated by the partial trace, give the dead state I/7I/7.

For a physicist: this is the analogue of renormalization-group invariance — the properties of a field theory do not depend on the scale of observation (up to running coupling constants). Here the coupling δ\delta between the parts plays the role of the running coupling: the corrections are of order δ\delta and are small only when δ\delta is.

For a biologist: the same principles of homeostasis can operate at the level of the cell, the organ and the organism wherever the parts are weakly coupled; the theorem does not say that they must.

Connection: section–retraction T-58′, frame rigidity, threshold robustness T-124d

Statement [T]

Let kk identical holons have the state σ∈D(C7)\sigma \in \mathcal{D}(\mathbb{C}^7), let ρk∈D(C7k)\rho_k \in \mathcal{D}(\mathbb{C}^{7^k}) be the state of the coupled collection, and let the aggregation be a CPTP channel Φk:D(C7k)→D(C7)\Phi_k: \mathcal{D}(\mathbb{C}^{7^k}) \to \mathcal{D}(\mathbb{C}^7). Assume

(AGG) (a) consistency: Φk(σ⊗k)=σ\Phi_k(\sigma^{\otimes k}) = \sigma — aggregating uncoupled copies returns the constituent (the partial trace and the mean of the single-copy marginals both qualify; the mean marginal Mk\mathcal{M}_k is the only permutation-invariant choice, Theorem 9.5 (a)); (b) weak coupling: dB(ρk,σ⊗k)≤δd_B(\rho_k, \sigma^{\otimes k}) \leq \delta — or, for Φk=Mk\Phi_k = \mathcal{M}_k, only (b′) max⁡i12∥(ρk)i−σ∥1≤δ\max_i \tfrac12\lVert (\rho_k)_i - \sigma \rVert_1 \leq \delta on the single-copy marginals, which (b) implies.

Then the aggregate Γ(k):=Φk(ρk)\Gamma^{(k)} := \Phi_k(\rho_k) satisfies dB(Γ(k),σ)≤δd_B(\Gamma^{(k)}, \sigma) \leq \delta (under (b)) and ∥Γ(k)−σ∥F≤2δ\|\Gamma^{(k)} - \sigma\|_F \leq 2\delta (under (b) or (b′)), and, with ΔP:=∣P(Γ(k))−P(σ)∣\Delta_P := |P(\Gamma^{(k)}) - P(\sigma)|:

  • ΔP≤4δP(σ)+4δ2\Delta_P \leq 4\delta\sqrt{P(\sigma)} + 4\delta^2 and ∣R(Γ(k))−R(σ)∣≤7ΔP|R(\Gamma^{(k)}) - R(\sigma)| \leq 7\Delta_P;
  • ∣Φ(Γ(k))−Φ(σ)∣≤7ΔP+196 P(σ) δ|\Phi(\Gamma^{(k)}) - \Phi(\sigma)| \leq 7\Delta_P + 196\,P(\sigma)\,\delta — a crude global bound; T-124d gives the first-order sensitivity;
  • ∣GapΓ(k)(i,j)−Gapσ(i,j)∣≤πδ/∣σij∣|\mathrm{Gap}_{\Gamma^{(k)}}(i,j) - \mathrm{Gap}_{\sigma}(i,j)| \leq \pi\delta/|\sigma_{ij}| whenever 2δ<∣σij∣2\delta < |\sigma_{ij}|;
  • each L2 condition P>2/7P > 2/7, R≥1/3R \geq 1/3, Φ≥1\Phi \geq 1 keeps its truth value when σ\sigma clears the threshold by more than the corresponding deviation.

Φ\Phi and Gap are compared in the fixed frame: they are frame-pinned, not G2G_2-invariant. No bound is claimed for the full L-level, which also involves DdiffD_{\mathrm{diff}} and, at L3, R(2)R^{(2)}.

Proof (5 steps).

Step 1 (Contractivity carries the conclusion because of (AGG a)). The fidelity does not decrease under CPTP maps, so the Bures distance does not increase (standard result):

dBures(Φk(ρ),Φk(ρ′))≤dBures(ρ,ρ′)d_{\mathrm{Bures}}(\Phi_k(\rho), \Phi_k(\rho')) \leq d_{\mathrm{Bures}}(\rho, \rho')

With ρ=ρk\rho = \rho_k, ρ′=σ⊗k\rho' = \sigma^{\otimes k} and (AGG a): dB(Γ(k),σ)≤dB(ρk,σ⊗k)≤δd_B(\Gamma^{(k)}, \sigma) \leq d_B(\rho_k, \sigma^{\otimes k}) \leq \delta. Without (a) the second image is not σ\sigma, and nothing about σ\sigma follows.

Step 2 (From Bures to trace and Frobenius norms). With dB2=2(1−F)d_B^2 = 2(1 - \sqrt{F}) one has 1−F=dB2−dB4/4≤dB21 - F = d_B^2 - d_B^4/4 \leq d_B^2, and the Fuchs–van de Graaf inequality 12∥ρ−ρ′∥1≤1−F\tfrac12\|\rho - \rho'\|_1 \leq \sqrt{1 - F} (arXiv:quant-ph/9712042) gives 12∥Γ(k)−σ∥1≤δ\tfrac12\|\Gamma^{(k)} - \sigma\|_1 \leq \delta. Under (b′) with Φk=Mk\Phi_k = \mathcal{M}_k the same bound follows from the convexity of the trace norm: 12∥Mk(ρk)−σ∥1≤max⁡i12∥(ρk)i−σ∥1\tfrac12\lVert \mathcal{M}_k(\rho_k) - \sigma \rVert_1 \leq \max_i \tfrac12\lVert (\rho_k)_i - \sigma \rVert_1; and (b) implies (b′), because the partial trace does not increase the trace distance. Since ∥X∥F≤∥X∥1\|X\|_F \leq \|X\|_1, the deviation X:=Γ(k)−σX := \Gamma^{(k)} - \sigma has ∥X∥F≤2δ\|X\|_F \leq 2\delta. Steps 3–5 use nothing else.

Step 3 (Purity and reflection). P(σ+X)−P(σ)=2 Tr(σX)+Tr(X2)P(\sigma + X) - P(\sigma) = 2\,\mathrm{Tr}(\sigma X) + \mathrm{Tr}(X^2), so by Cauchy–Schwarz ΔP≤2∥σ∥F∥X∥F+∥X∥F2≤4δP(σ)+4δ2\Delta_P \leq 2\|\sigma\|_F\|X\|_F + \|X\|_F^2 \leq 4\delta\sqrt{P(\sigma)} + 4\delta^2 (Bound 1 of T-124d). For R=1/(7P)R = 1/(7P): ∣ΔR∣=ΔP/(7PP′)≤7ΔP|\Delta R| = \Delta_P/(7PP') \leq 7\Delta_P, because P,P′≥1/7P, P' \geq 1/7.

Step 4 (Integration and Gap, in the fixed frame). Write Φ=P/D−1\Phi = P/D - 1 with D=∑iγii2≥1/7D = \sum_i \gamma_{ii}^2 \geq 1/7 (Cauchy–Schwarz, since ∑iγii=1\sum_i \gamma_{ii} = 1). Then ∣ΔΦ∣≤ΔP/D′+P ∣ΔD∣/(DD′)≤7ΔP+49 P(σ) ∣ΔD∣|\Delta\Phi| \leq \Delta_P/D' + P\,|\Delta D|/(DD') \leq 7\Delta_P + 49\,P(\sigma)\,|\Delta D|, and ∣ΔD∣≤(D+D′) ∥X∥F≤4δ|\Delta D| \leq (\sqrt{D} + \sqrt{D'})\,\|X\|_F \leq 4\delta. For Gap(i,j)=∣sin⁡(arg⁡γij)∣\mathrm{Gap}(i,j) = |\sin(\arg\gamma_{ij})|: if ∣Xij∣<∣σij∣|X_{ij}| < |\sigma_{ij}|, the phase of σij+Xij\sigma_{ij} + X_{ij} differs from that of σij\sigma_{ij} by at most arcsin⁡(∣Xij∣/∣σij∣)≤π2∣Xij∣/∣σij∣\arcsin(|X_{ij}|/|\sigma_{ij}|) \leq \tfrac{\pi}{2}|X_{ij}|/|\sigma_{ij}|, and ∣sin⁡∣|\sin| is 1-Lipschitz; with ∣Xij∣≤∥X∥F≤2δ|X_{ij}| \leq \|X\|_F \leq 2\delta this is the stated bound. The comparison is meaningful because (AGG a) holds as a matrix identity in one frame; a G2G_2 rotation of either state would change Φ\Phi and Gap.

Step 5 (Thresholds). If P(σ)−2/7P(\sigma) - 2/7 exceeds the bound on ΔP\Delta_P, then P(Γ(k))>2/7P(\Gamma^{(k)}) > 2/7 as well; likewise for R≥1/3R \geq 1/3 and Φ≥1\Phi \geq 1. A state that lies within the deviation of a threshold can cross it in either direction. ■\blacksquare

Corollary 9.2a ((AGG) at the stationary state of weakly coupled holons) [T]

Let kk identical embodied holons with an anchor outside the null set of Theorem 9.4 be coupled by −ig Hint-ig\,H_{\mathrm{int}}, and let the aggregation be the partial trace onto one holon or the mean of the single-copy marginals. For ∣g∣|g| small the stationary state ρk=X(g)\rho_k = X(g) near σ⊗k\sigma^{\otimes k} satisfies (AGG): (a) holds by the choice of aggregation, and (b) holds in trace norm with δ=12∥X(g)−σ⊗k∥1=O(g)\delta = \tfrac12\lVert X(g) - \sigma^{\otimes k}\rVert_1 = O(g). The conclusions of Theorem 9.2 therefore hold at the stationary state with deviations O(g)O(g).

Proof. The branch X(g)X(g) and its derivative are those of Corollary 9.1a, now for kk factors. Both aggregations are CPTP and return σ\sigma on σ⊗k\sigma^{\otimes k}, so 12∥Γ(k)−σ∥1≤δ\tfrac12\lVert\Gamma^{(k)} - \sigma\rVert_1 \leq \delta; steps 2–5 of Theorem 9.2 use only this bound, ∥Γ(k)−σ∥F≤2δ\lVert\Gamma^{(k)} - \sigma\rVert_F \leq 2\delta. ■\blacksquare Witness (test_non_degeneracy_is_generic_and_aggregation_follows_from_weak_coupling): two identical embodied holons with a generic coupling X⊗YX \otimes Y of unit norm; ∥X(g)−σ⊗σ∥1/g=0.13421\lVert X(g) - \sigma \otimes \sigma\rVert_1 / g = 0.13421 at g=0.01g = 0.01 and 0.134190.13419 at g=0.02g = 0.02.

What does not need (AGG). If the aggregate is itself a holon — assumption (HOL) of Theorem 9.1 — that theorem gives it its own non-trivial attractor, P(ρ∗(k))>1/7P(\rho_*^{(k)}) > 1/7 (T-96 [T]). That statement concerns the aggregate's own dynamics, not how its invariants compare with those of its parts.

Corollary (Fractal structure) [T at weak coupling]

Scale invariance and fractal closure, both proved for weak coupling through the canonical aggregation (Theorem 9.5), give UHM a fractal structure at every scale at which the constituents are embodied, viable and weakly coupled, ∣g∣ s(Hint)<εV|g|\,s(H_{\mathrm{int}}) < \varepsilon_V: the canonical aggregate is viable and its invariants lie within O(g)O(g) of a part's. What is inherited is the parts' state, not something new: the canonical aggregate depends only on the marginals. Where the coupling is strong, nothing of the kind need hold — the aggregate of two viable holons can be I/7I/7 (Theorem 9.6); a composite that is assumed to be a holon in the sense of (HOL) still has its own attractor.


A joint state can inherit the marginal properties without being correlated. For two subsystems, I(1:2)=0I(1{:}2)=0 exactly for a product of its own margins. Which couplings take the steady state outside this product set?

Theorem 9.3 (CC-7: Emergence) [T for almost every anchor]​

An interaction commuting with the product anchor can leave it stationary. A local drive may change one marginal without introducing correlations. Hence the comparison is to the product of the current margins. Both cases are realized below. Numerical witness: test_coupled_holons_can_have_a_product_stationary_state in website/scripts/check_core_numbers.py.

In Plain Terms

Two pendulums hung from one beam swing in step because the beam passes motion from one to the other; two pendulums whose coupling acts only on what each is already doing stay independent, however strong the coupling. Theorem 9.3 says which kind a coupling between two holons is. The whole acquires information that is not in the parts — mutual information I>0I > 0 — exactly when the coupling has a correlating part at the parts' own steady states; a coupling that commutes with those states, or acts on one holon alone, leaves the pair uncorrelated.

For a psychologist: two people who interact are not thereby correlated; the interaction has to depend jointly on what each of them is doing. For a physicist: this is the familiar statement that a weak perturbation correlates two subsystems at first order only through the part of [Hint,ρ1⊗ρ2][H_{\mathrm{int}}, \rho_1 \otimes \rho_2] that is not a sum of local terms.

Connection: Quantum mutual information, canonical extension of R\mathcal{R} to composite systems

Setting. Each holon i=1,2i = 1, 2 has the generator of evolution, Li[Γ]=−i[Hi,Γ]+DΩ[Γ]+κ(Γ) gV(P) (φcoh(Γ)−Γ)+μ (σi−Γ)\mathcal{L}_i[\Gamma] = -i[H_i, \Gamma] + \mathcal{D}_\Omega[\Gamma] + \kappa(\Gamma)\, g_V(P)\,(\varphi_{\mathrm{coh}}(\Gamma) - \Gamma) + \mu\,(\sigma_i - \Gamma), where the last term is the backbone injection toward the anchor σi=π(B(x))\sigma_i = \pi(\mathcal{B}(x)) of an embodied holon (T-148), μ>0\mu > 0. Freezing the scalars κ\kappa, gVg_V and k=1−Rk = 1 - R at a state gg gives a linear generator Mi(g)\mathcal{M}_i^{(g)} of a CPTP semigroup with Li[Γ]=Mi(Γ)(Γ)\mathcal{L}_i[\Gamma] = \mathcal{M}_i^{(\Gamma)}(\Gamma). The composite has the canonical extension plus a Hamiltonian coupling:

L(12)[X]=(M1(X1)⊗id)(X)+(id⊗M2(X2))(X)−i g [Hint,X],X1=Tr2X,  X2=Tr1X.\mathcal{L}^{(12)}[X] = (\mathcal{M}_1^{(X_1)} \otimes \mathrm{id})(X) + (\mathrm{id} \otimes \mathcal{M}_2^{(X_2)})(X) - i\,g\,[H_{\mathrm{int}}, X], \qquad X_1 = \mathrm{Tr}_2 X,\; X_2 = \mathrm{Tr}_1 X .

Let Li[ρ∗(i)]=0\mathcal{L}_i[\rho_*^{(i)}] = 0, σ:=ρ∗(1)⊗ρ∗(2)\sigma := \rho_*^{(1)} \otimes \rho_*^{(2)}, and let Πc(s1⊗s2)(Y):=Y−Tr2Y⊗s2−s1⊗Tr1Y+Tr(Y) s1⊗s2\Pi_c^{(s_1 \otimes s_2)}(Y) := Y - \mathrm{Tr}_2 Y \otimes s_2 - s_1 \otimes \mathrm{Tr}_1 Y + \mathrm{Tr}(Y)\, s_1 \otimes s_2 be the projection onto the correlation part: it vanishes exactly on operators of the form a⊗s2+s1⊗ba \otimes s_2 + s_1 \otimes b.

(ND) Non-degeneracy: each ρ∗(i)\rho_*^{(i)} is a non-degenerate fixed point — the Jacobian of Li\mathcal{L}_i at ρ∗(i)\rho_*^{(i)} is invertible on traceless Hermitian operators, and P(ρ∗(i))∉{2/7,3/7}P(\rho_*^{(i)}) \notin \{2/7, 3/7\}, so that the gate gVg_V is differentiable there.

Theorem 9.3 (CC-7) [T for almost every anchor]

(i) Exact, any gg. σ\sigma is a stationary state of the coupled composite if and only if [Hint,σ]=0[H_{\mathrm{int}}, \sigma] = 0. In that case the pair has a stationary state with I(H1:H2)=0I(\mathbb{H}_1 : \mathbb{H}_2) = 0 at every coupling strength.

(ii) Exact, any gg. A product s1⊗s2s_1 \otimes s_2 is a stationary state if and only if Πc(s1⊗s2)([Hint,s1⊗s2])=0\Pi_c^{(s_1 \otimes s_2)}\bigl([H_{\mathrm{int}}, s_1 \otimes s_2]\bigr) = 0 and each sis_i is stationary for Li−ig[Himf,⋅ ]\mathcal{L}_i - i g [H_i^{\mathrm{mf}}, \cdot\,] with the mean fields H1mf=Tr2[(I⊗s2)Hint]H_1^{\mathrm{mf}} = \mathrm{Tr}_2[(I \otimes s_2) H_{\mathrm{int}}], H2mf=Tr1[(s1⊗I)Hint]H_2^{\mathrm{mf}} = \mathrm{Tr}_1[(s_1 \otimes I) H_{\mathrm{int}}]. A stationary state of the pair is uncorrelated exactly when it is such a product; in particular a local coupling HA⊗I+I⊗HBH_A \otimes I + I \otimes H_B never correlates the pair.

(iii) Weak coupling, under (ND) — which holds for every pair of anchors outside a closed null set (Theorem 9.4). For ∣g∣|g| small there is a unique stationary state X(g)X(g) near σ\sigma, smooth in gg, and

X(g)−X1(g)⊗X2(g)=g C1+O(g2),C1=Jc−1 Πc(σ)(i[Hint,σ]),X(g) - X_1(g) \otimes X_2(g) = g\, C_1 + O(g^2), \qquad C_1 = \mathcal{J}_c^{-1}\, \Pi_c^{(\sigma)}\bigl(i[H_{\mathrm{int}}, \sigma]\bigr),

where Jc=M1(ρ∗(1))⊗id+id⊗M2(ρ∗(2))\mathcal{J}_c = \mathcal{M}_1^{(\rho_*^{(1)})} \otimes \mathrm{id} + \mathrm{id} \otimes \mathcal{M}_2^{(\rho_*^{(2)})} restricted to the correlation space, whose spectrum lies in Re λ≤−2μ\mathrm{Re}\,\lambda \leq -2\mu. Hence, if Πc(σ)([Hint,σ])≠0\Pi_c^{(\sigma)}([H_{\mathrm{int}}, \sigma]) \neq 0, then I(X(g))≥12∥X(g)−X1(g)⊗X2(g)∥12>0I(X(g)) \geq \tfrac12 \lVert X(g) - X_1(g) \otimes X_2(g) \rVert_1^2 > 0 for every small g≠0g \neq 0, and I=Θ(g2)I = \Theta(g^2); if Πc(σ)([Hint,σ])=0\Pi_c^{(\sigma)}([H_{\mathrm{int}}, \sigma]) = 0, the correlation is at most O(g2)O(g^2).

Proof.

Step 1 (Product states). The extension acts on a product through its factors: (M⊗id)(a⊗b)=M(a)⊗b(\mathcal{M} \otimes \mathrm{id})(a \otimes b) = \mathcal{M}(a) \otimes b, and the scalars of Mi(Xi)\mathcal{M}_i^{(X_i)} are read on the marginals, which for s1⊗s2s_1 \otimes s_2 are s1,s2s_1, s_2. Hence

L(12)[s1⊗s2]=L1[s1]⊗s2+s1⊗L2[s2]−ig [Hint,s1⊗s2].\mathcal{L}^{(12)}[s_1 \otimes s_2] = \mathcal{L}_1[s_1] \otimes s_2 + s_1 \otimes \mathcal{L}_2[s_2] - i g\, [H_{\mathrm{int}}, s_1 \otimes s_2] .

For si=ρ∗(i)s_i = \rho_*^{(i)} the first two terms vanish, which proves (i). For (ii): Tr2[Hint,s1⊗s2]=[H1mf,s1]\mathrm{Tr}_2 [H_{\mathrm{int}}, s_1 \otimes s_2] = [H_1^{\mathrm{mf}}, s_1] (cyclicity of the partial trace in the second factor), and symmetrically for Tr1\mathrm{Tr}_1, so the two partial traces of the equation are the two mean-field equations; Πc\Pi_c annihilates the local terms L1[s1]⊗s2\mathcal{L}_1[s_1] \otimes s_2 and s1⊗L2[s2]s_1 \otimes \mathcal{L}_2[s_2], so what remains is Πc([Hint,s1⊗s2])=0\Pi_c([H_{\mathrm{int}}, s_1 \otimes s_2]) = 0. The three conditions together are equivalent to the equation, because Y=Πc(Y)+Tr2Y⊗s2+s1⊗Tr1Y−Tr(Y) s1⊗s2Y = \Pi_c(Y) + \mathrm{Tr}_2 Y \otimes s_2 + s_1 \otimes \mathrm{Tr}_1 Y - \mathrm{Tr}(Y)\, s_1 \otimes s_2. For Hint=HA⊗I+I⊗HBH_{\mathrm{int}} = H_A \otimes I + I \otimes H_B the commutator with a product is local, and Πc\Pi_c of it is zero. Finally I(X)=D(X ∥ X1⊗X2)I(X) = D(X \,\|\, X_1 \otimes X_2) vanishes exactly on products.

Step 2 (The Jacobian splits). Let J\mathcal{J} be the Jacobian of L(12)\mathcal{L}^{(12)} at g=0g = 0, X=σX = \sigma, on traceless Hermitian operators. Along a local direction x⊗ρ∗(2)x \otimes \rho_*^{(2)} only the first marginal moves, and J(x⊗ρ∗(2))=(DL1 x)⊗ρ∗(2)+x⊗L2[ρ∗(2)]=(DL1 x)⊗ρ∗(2)\mathcal{J}(x \otimes \rho_*^{(2)}) = (D\mathcal{L}_1\, x) \otimes \rho_*^{(2)} + x \otimes \mathcal{L}_2[\rho_*^{(2)}] = (D\mathcal{L}_1\, x) \otimes \rho_*^{(2)} — the second factor's frozen generator applied to its own fixed point gives zero. Along a correlation direction (Tr1Y=Tr2Y=0\mathrm{Tr}_1 Y = \mathrm{Tr}_2 Y = 0) neither marginal moves, the scalars stay frozen, and JY=JcY\mathcal{J} Y = \mathcal{J}_c Y, which again has zero partial traces because the Mi\mathcal{M}_i preserve the trace. So J\mathcal{J} is block-diagonal: the Jacobians of the two holons on the local blocks, Jc\mathcal{J}_c on the correlation block; and it commutes with Πc(σ)\Pi_c^{(\sigma)}.

Step 3 (The correlation block is invertible). On traceless YY the backbone term acts as −μY-\mu Y (the replacement Y↦Tr(Y)σiY \mapsto \mathrm{Tr}(Y)\sigma_i annihilates it), and the rest of Mi\mathcal{M}_i generates a trace-preserving CPTP semigroup that maps traceless operators to traceless ones; hence ∥etMiY∥1≤e−μt∥Y∥1\lVert e^{t\mathcal{M}_i} Y \rVert_1 \leq e^{-\mu t} \lVert Y \rVert_1 and the spectrum of Mi\mathcal{M}_i on traceless operators has Re λ≤−μ\mathrm{Re}\,\lambda \leq -\mu. The spectrum of Jc\mathcal{J}_c consists of the sums λ+λ′\lambda + \lambda' of such eigenvalues: Re≤−2μ<0\mathrm{Re} \leq -2\mu < 0. No assumption is needed here; the backbone of an embodied holon supplies it.

Step 4 (First order). Under (ND) the local blocks are invertible too, so J\mathcal{J} is, and the implicit function theorem gives the branch X(g)X(g). Differentiating L(12)[X(g)]=0\mathcal{L}^{(12)}[X(g)] = 0 at g=0g = 0: JX′(0)=i[Hint,σ]\mathcal{J} X'(0) = i[H_{\mathrm{int}}, \sigma]; applying Πc(σ)\Pi_c^{(\sigma)}, which commutes with J\mathcal{J}, gives Πc(σ)X′(0)=C1\Pi_c^{(\sigma)} X'(0) = C_1. Since X−X1⊗X2=Πc(σ)(X)−(X1−ρ∗(1))⊗(X2−ρ∗(2))X - X_1 \otimes X_2 = \Pi_c^{(\sigma)}(X) - (X_1 - \rho_*^{(1)}) \otimes (X_2 - \rho_*^{(2)}), the correlation is gC1+O(g2)g C_1 + O(g^2), and C1≠0C_1 \neq 0 exactly when Πc(σ)([Hint,σ])≠0\Pi_c^{(\sigma)}([H_{\mathrm{int}}, \sigma]) \neq 0. The bound on II is the quantum Pinsker inequality D(ρ ∥ τ)≥12∥ρ−τ∥12D(\rho \,\|\, \tau) \geq \tfrac12 \lVert \rho - \tau \rVert_1^2. ■\blacksquare

Status: [T] for almost every anchor. Parts (i) and (ii) use nothing beyond the form of the composite generator and hold unconditionally. Part (iii) needs (ND), and Theorem 9.4 below proves it for every pair of anchors outside a closed Lebesgue-null set. It also needs the attractor to exist. An embodied holon always has a stationary state: its flow maps the compact convex set of states into itself, so each time-tt map has a fixed point (Brouwer), and a limit of such points as t→0t \to 0 is stationary. An isolated holon with the self-registering φs\varphi_s has seven non-degenerate ones for small HH (evolution), and there the correlation block is invertible too, with Re≤−2κgVR\mathrm{Re} \leq -2\kappa g_V R in place of −2μ-2\mu (the anchor term acts as −κgVR-\kappa g_V R on traceless operators). Without an anchor of either kind there is nothing to apply (iii) to: an isolated holon with the canonical φcoh\varphi_{\mathrm{coh}} (anchor I/7I/7) has none besides I/7I/7, since Tr(Γφcoh(Γ))≤P−(P−1/7)/(7P)\mathrm{Tr}(\Gamma \varphi_{\mathrm{coh}}(\Gamma)) \leq P - (P - 1/7)/(7P), so regeneration and dissipation both lower PP (over 100 random states the left side minus the right is at most −0.020-0.020; twenty pure starts all end at P=0.14286P = 0.14286 by t=60t = 60); the anchor σi\sigma_i of an embodied holon is what gives it one (T-148).

Numerical check (test_coupled_holons_can_have_a_product_stationary_state). Two embodied holons with the canonical ingredients above (α=1/2\alpha = 1/2, μ=1\mu = 1, κ=1/7+CohE\kappa = 1/7 + \mathrm{Coh}_E, gV=clamp(7P−2,0,1)g_V = \mathrm{clamp}(7P - 2, 0, 1), random HiH_i of norm scale 0.30.3, anchors of purity weight 0.850.85 and 0.80.8) have attractors with P=0.362P = 0.362 and 0.3030.303 (residual below 10−1510^{-15}). All couplings are normalised to operator norm 0.30.3.

  • (i) Hint∝(ρ∗(1)−I/7)⊗(ρ∗(2)−I/7)H_{\mathrm{int}} \propto (\rho_*^{(1)} - I/7) \otimes (\rho_*^{(2)} - I/7): ∥[Hint,σ]∥F=1.7×10−17\lVert [H_{\mathrm{int}}, \sigma] \rVert_F = 1.7 \times 10^{-17}. From a random state on C49\mathbb{C}^{49} the flow reaches σ\sigma to 4.9×10−164.9 \times 10^{-16} by t=24t = 24; ∣I∣<10−15\lvert I \rvert < 10^{-15}.
  • (ii) Hint=HA⊗IH_{\mathrm{int}} = H_A \otimes I: ∥[Hint,σ]∥F=0.052\lVert [H_{\mathrm{int}}, \sigma] \rVert_F = 0.052; the stationary state moves 0.0270.027 away from σ\sigma and stays a product to 4×10−164 \times 10^{-16}; ∣I∣<10−15\lvert I \rvert < 10^{-15}.
  • A generic X⊗YX \otimes Y: correlation ∥X−X1⊗X2∥F=0.013\lVert X - X_1 \otimes X_2 \rVert_F = 0.013, I=2.1×10−3I = 2.1 \times 10^{-3}.
  • (iii) In a run of the same model with κ0=ω0∣γOE∣∣γOU∣/γOO\kappa_0 = \omega_0 \lvert\gamma_{OE}\rvert \lvert\gamma_{OU}\rvert / \gamma_{OO}: the single-holon Jacobians have spectra with Re λ≤−1.30\mathrm{Re}\,\lambda \leq -1.30 and −1.42-1.42, so (ND) holds for this pair; under the coupling of (i), three random starts on C49\mathbb{C}^{49} end within 10−1310^{-13} of σ\sigma; for a generic X⊗YX \otimes Y of unit norm at g=0.02g = 0.02 and 0.040.04, the measured correlation matches gC1g C_1 to relative 5.8×10−35.8 \times 10^{-3} and 1.16×10−21.16 \times 10^{-2} (error linear in gg), and I/g2=0.02437I/g^2 = 0.02437 at both.

Theorem 9.4 (Non-degeneracy is generic) [T]​

Theorem 9.4 [T]

Let a holon be embodied, with generator Lσ[Γ]=−i[H,Γ]+DΩ[Γ]+κ(Γ) gV(P) (φ(Γ)−Γ)+μ(σ−Γ)\mathcal{L}_\sigma[\Gamma] = -i[H,\Gamma] + \mathcal{D}_\Omega[\Gamma] + \kappa(\Gamma)\,g_V(P)\,(\varphi(\Gamma) - \Gamma) + \mu(\sigma - \Gamma), μ>0\mu > 0, κ\kappa smooth, φ=φcoh\varphi = \varphi_{\mathrm{coh}} or φs\varphi_s, and a full-rank anchor σ\sigma. There is a closed Lebesgue-null set NN of anchors such that for σ∉N\sigma \notin N every stationary state of Lσ\mathcal{L}_\sigma is non-degenerate and has P∉{2/7,3/7}P \notin \{2/7, 3/7\}. Its complement is open and dense. Hence (ND) holds for every pair of anchors outside N×NN \times N.

Proof. On the open set UU of trace-one Hermitian matrices with P∉{2/7,3/7}P \notin \{2/7, 3/7\} the map F(Γ,σ)=Lσ[Γ]F(\Gamma, \sigma) = \mathcal{L}_\sigma[\Gamma] is smooth (P≥1/7P \geq 1/7 on trace-one Hermitian matrices, so RR and Γ2/P\Gamma^2/P are smooth), with values in the traceless Hermitian matrices. Its derivative in σ\sigma is μ\mu times the identity on traceless directions, which is onto; so 00 is a regular value of FF on U×{σ full rank}U \times \{\sigma \text{ full rank}\}. By the parametric transversality theorem (V. Guillemin, A. Pollack, Differential Topology, Prentice-Hall 1974, Ch. 2 §3), for almost every σ\sigma the value 00 is regular for F(⋅,σ)F(\cdot, \sigma) on UU: every zero there is non-degenerate. On each kink surface Kc={P=c}K_c = \{P = c\}, c∈{2/7,3/7}c \in \{2/7, 3/7\}, take the one-sided smooth continuation of gVg_V (it agrees with gVg_V on KcK_c); its restriction to KcK_c, a 47-dimensional manifold, is again a submersion in σ\sigma, and transversality to 00 in a 48-dimensional space means no zeros, so for almost every σ\sigma no stationary state lies on KcK_c. The bad set NN is closed: a limit of anchors with a degenerate or kink stationary state has one too, since stationary states lie in the compact set of states. A closed null set has a dense open complement. ■\blacksquare

Witness (test_non_degeneracy_is_generic_and_aggregation_follows_from_weak_coupling): 12 embodied holons with random HH of scale 0.30.3 and random anchors of pure weight 0.60.6–0.90.9; the attractors have PP from 0.2260.226 to 0.3410.341, at least 1.3⋅10−31.3 \cdot 10^{-3} from 2/72/7 and 3/73/7, and Jacobians with min⁡∣Re λ∣\min\lvert\mathrm{Re}\,\lambda\rvert from 1.191.19 to 1.491.49 — all non-degenerate.

What remains of "emergence". For a correlated joint state the marginals do not determine it, and I=S(ρ1)+S(ρ2)−S(ρ12)>0I = S(\rho_1) + S(\rho_2) - S(\rho_{12}) > 0 is the information the partial traces discard — a standard identity, true of every correlated pair, coupled thermostats included. Theorem 9.3 says when the dynamics of coupled holons produces such a state; it does not say that interaction alone does.

Connection to Löwer incompleteness

When I>0I > 0 — under the criterion of (iii), not for every coupling — subsystem H1\mathbb{H}_1 cannot reconstruct the joint state from ρ1\rho_1 alone (observation-fibre theorem).

Theorem 9.5 (Canonical aggregation: viability and invariants pass to the aggregate at weak coupling) [T at weak coupling]​

Theorems 9.1 and 9.2 needed two things the corpus did not have: a map from the composite's states on (C7)⊗k(\mathbb{C}^7)^{\otimes k} to D(C7)\mathcal{D}(\mathbb{C}^7), and a reason why the image of a coupled composite should be a living holon. The theorem below supplies both. The map is not chosen: it is the only one that treats the parts alike and returns a part's state when the parts are uncoupled. That it sends coupled composites to living holons is proved for weak coupling, with an explicit threshold, and Theorem 9.6 shows that the threshold cannot be dropped.

In Plain Terms

If each of kk musicians plays in tune, and they listen to each other only a little, the ensemble — heard as "one musician", by averaging what each plays — is also in tune; how far it can drift is fixed by how strongly they couple, and the listening can only push each musician by that much. If they lock together strongly enough, each musician's own line can dissolve into pure harmony with the others, and the averaged "one musician" is then noise, though the ensemble as a whole is ordered.

Setting. kk embodied holons with generators Li[Γ]=−i[Hi,Γ]+DΩ[Γ]+κi(Γ) gV(P) (φi(Γ)−Γ)+μi(σi−Γ)\mathcal{L}_i[\Gamma] = -i[H_i, \Gamma] + \mathcal{D}_\Omega[\Gamma] + \kappa_i(\Gamma)\,g_V(P)\,(\varphi_i(\Gamma) - \Gamma) + \mu_i(\sigma_i - \Gamma), μi>0\mu_i > 0, any regeneration target φi\varphi_i and any κi≥0\kappa_i \geq 0, with the gate gV(P)=clamp((P−2/7)/(3/7−2/7),0,1)g_V(P) = \mathrm{clamp}\bigl((P - 2/7)/(3/7 - 2/7), 0, 1\bigr) of evolution, which vanishes for P≤2/7P \leq 2/7. The composite on (C7)⊗k(\mathbb{C}^7)^{\otimes k} carries the canonical extension of Theorem 9.3 and a coupling: L(k)[X]=∑iMi(Xi)\mathcal{L}^{(k)}[X] = \sum_i \mathcal{M}_i^{(X_i)} acting on factor ii, minus ig[Hint,X]ig[H_{\mathrm{int}}, X], with marginals Xi=Tr≠iXX_i = \mathrm{Tr}_{\neq i} X. Write s(H)=λmax⁡(H)−λmin⁡(H)s(H) = \lambda_{\max}(H) - \lambda_{\min}(H) for the spread of a Hermitian operator, and ρlin(i)\rho_{\mathrm{lin}}^{(i)} for the stationary state of the regeneration-free part Li0=−i[Hi,⋅]+DΩ+μi(σiTr−id)\mathcal{L}_i^0 = -i[H_i, \cdot] + \mathcal{D}_\Omega + \mu_i(\sigma_i \mathrm{Tr} - \mathrm{id}). Call holon ii viable if every stationary state of Li\mathcal{L}_i has P>2/7P > 2/7.

Theorem 9.5 [T at weak coupling]

(a) The canonical aggregation [T]. Among all linear maps AA from operators on (C7)⊗k(\mathbb{C}^7)^{\otimes k} to operators on C7\mathbb{C}^7 that are invariant under permutations of the factors and consistent on uncoupled identical copies — A(σ⊗k)=σA(\sigma^{\otimes k}) = \sigma for every state σ\sigma — there is exactly one, the mean marginal

Mk(X)=1k∑i=1kTr≠i X.\mathcal{M}_k(X) = \frac1k \sum_{i=1}^k \mathrm{Tr}_{\neq i}\, X .

It is CPTP, U(7)U(7)-covariant (Mk(U⊗kXU†⊗k)=UMk(X)U†\mathcal{M}_k(U^{\otimes k} X U^{\dagger\otimes k}) = U \mathcal{M}_k(X) U^\dagger, so in particular G2G_2-covariant), and it depends on XX only through the marginals.

(b) Exact marginal equation [T], any gg. Along every trajectory of the composite,

dXidt=Li[Xi]−ig Tr≠i[Hint,X],∥Tr≠i[Hint,X]∥1≤s(Hint).\frac{d X_i}{dt} = \mathcal{L}_i[X_i] - ig\,\mathrm{Tr}_{\neq i}[H_{\mathrm{int}}, X], \qquad \lVert \mathrm{Tr}_{\neq i}[H_{\mathrm{int}}, X] \rVert_1 \leq s(H_{\mathrm{int}}) .

(c) Viability of a part is a threshold on its linear part [T]. Holon ii is viable if and only if P(ρlin(i))>2/7P(\rho_{\mathrm{lin}}^{(i)}) > 2/7. In that case every state Γ\Gamma with P(Γ)≤2/7P(\Gamma) \leq 2/7 has ∥Li[Γ]∥1≥εV(i):=μi(P(ρlin(i))−2/7)/(2P(ρlin(i)))\lVert \mathcal{L}_i[\Gamma] \rVert_1 \geq \varepsilon_V^{(i)} := \mu_i \bigl(P(\rho_{\mathrm{lin}}^{(i)}) - 2/7\bigr) / \bigl(2\sqrt{P(\rho_{\mathrm{lin}}^{(i)})}\bigr). A sufficient condition in terms of the anchor: μi (P(σi)−2/7)>2P(σi) ∥−i[Hi,σi]+DΩ[σi]∥1\mu_i\,(P(\sigma_i) - 2/7) > 2\sqrt{P(\sigma_i)}\,\lVert -i[H_i, \sigma_i] + \mathcal{D}_\Omega[\sigma_i] \rVert_1.

(d) Every stationary composite has living parts [T]. If every part is viable and ∣g∣ s(Hint)<min⁡iεV(i)|g|\,s(H_{\mathrm{int}}) < \min_i \varepsilon_V^{(i)}, then every stationary state of the composite — there is at least one — has P(Xi)>2/7P(X_i) > 2/7 for every ii. For identical parts and a permutation-symmetric stationary state, Mk(X)=X1\mathcal{M}_k(X) = X_1 is viable. Neither (ND) nor (HOL) is used.

(e) Every trajectory, with explicit constants, under backbone dominance [T]. If the parts are identical and μ>LR\mu > L_{\mathcal{R}} (the regime of backbone dominance, which gives a unique stationary state ρ∗\rho_*), then for every initial state of the composite and every t≥0t \geq 0

∥Xi(t)−ρ∗∥1≤e−(μ−LR)t ∥Xi(0)−ρ∗∥1+∣g∣ s(Hint)μ−LR,\lVert X_i(t) - \rho_* \rVert_1 \leq e^{-(\mu - L_{\mathcal{R}})t}\,\lVert X_i(0) - \rho_* \rVert_1 + \frac{|g|\,s(H_{\mathrm{int}})}{\mu - L_{\mathcal{R}}},

so 12∥Mk(X(t))−ρ∗∥1\tfrac12\lVert \mathcal{M}_k(X(t)) - \rho_* \rVert_1 obeys the same bound halved.

(f) The basin, in general [T]. If the parts are identical and ρ∗\rho_* is a non-degenerate stationary state of L\mathcal{L} whose Jacobian spectrum lies in Re λ<0\mathrm{Re}\,\lambda < 0, with basin of attraction B\mathfrak{B}, then for every compact K⊂BK \subset \mathfrak{B} there are g0,C,T>0g_0, C, T > 0 such that for ∣g∣<g0|g| < g_0 every trajectory of the composite whose initial marginals lie in KK satisfies ∥Xi(t)−ρ∗∥1≤C∣g∣\lVert X_i(t) - \rho_* \rVert_1 \leq C|g| for all t≥Tt \geq T.

(g) (HOL) up to a forcing of size ∣g∣ s(Hint)|g|\,s(H_{\mathrm{int}}) [T]. For identical parts, a permutation-invariant HintH_{\mathrm{int}} and a permutation-invariant initial state, the aggregate Γ(t)=Mk(X(t))\Gamma(t) = \mathcal{M}_k(X(t)) equals X1(t)X_1(t) and obeys Γ˙=L[Γ]+u(t)\dot\Gamma = \mathcal{L}[\Gamma] + u(t) with ∥u(t)∥1≤∣g∣ s(Hint)\lVert u(t) \rVert_1 \leq |g|\,s(H_{\mathrm{int}}): the canonical aggregate is a holon of the parts' own kind, driven by a bounded forcing. For a local coupling HA⊗I+I⊗HBH_A \otimes I + I \otimes H_B a product state stays a product and the forcing is the mean-field Hamiltonian term of Theorem 9.3 (ii): there (HOL) holds exactly.

Proof.

(a) On states, A(σ⊗k)=σ=σ (Tr σ)k−1A(\sigma^{\otimes k}) = \sigma = \sigma\,(\mathrm{Tr}\,\sigma)^{k-1}. Both sides are homogeneous polynomials of degree kk on the real space of Hermitian matrices that agree on the open cone of positive definite matrices, hence everywhere. Polarisation gives, for Hermitian a1,…,aka_1, \ldots, a_k and their symmetrised product Sym(a1⊗⋯⊗ak)\mathrm{Sym}(a_1 \otimes \cdots \otimes a_k), A(Sym(a1⊗⋯⊗ak))=1k∑iai∏j≠iTr aj=Mk(Sym(a1⊗⋯⊗ak))A(\mathrm{Sym}(a_1 \otimes \cdots \otimes a_k)) = \tfrac1k \sum_i a_i \prod_{j \neq i} \mathrm{Tr}\,a_j = \mathcal{M}_k(\mathrm{Sym}(a_1 \otimes \cdots \otimes a_k)). These symmetrised products span the permutation-invariant operators, and both AA and Mk\mathcal{M}_k factor through the symmetrisation X↦1k!∑πUπXUπ†X \mapsto \tfrac1{k!}\sum_\pi U_\pi X U_\pi^\dagger (AA by assumption, Mk\mathcal{M}_k because permuting the factors permutes the marginals). So A=MkA = \mathcal{M}_k. Partial traces are CPTP and Tr≠i(U⊗kXU†⊗k)=UXiU†\mathrm{Tr}_{\neq i}(U^{\otimes k} X U^{\dagger\otimes k}) = U X_i U^\dagger. Without the permutation invariance there is no uniqueness: each Tr≠i\mathrm{Tr}_{\neq i} alone is consistent.

(b) Mj(Xj)\mathcal{M}_j^{(X_j)} generates a trace-preserving semigroup, so Tr∘Mj(Xj)=0\mathrm{Tr} \circ \mathcal{M}_j^{(X_j)} = 0 and the partial trace over factor j≠ij \neq i kills the jj-th term; the ii-th term commutes with tracing out the other factors and gives Mi(Xi)(Xi)=Li[Xi]\mathcal{M}_i^{(X_i)}(X_i) = \mathcal{L}_i[X_i]. For the bound, [H,X]=[H−cI,X][H, X] = [H - cI, X] with c=(λmax⁡+λmin⁡)/2c = (\lambda_{\max} + \lambda_{\min})/2, so ∥[H,X]∥1≤2∥H−cI∥∞∥X∥1=s(H)\lVert [H, X] \rVert_1 \leq 2\lVert H - cI \rVert_\infty \lVert X \rVert_1 = s(H), and the partial trace does not increase the trace norm.

(c) For P(Γ)≤2/7P(\Gamma) \leq 2/7 the gate is closed and Li[Γ]=Li0[Γ]\mathcal{L}_i[\Gamma] = \mathcal{L}_i^0[\Gamma]. On traceless YY the anchor term reduces to −μiY-\mu_i Y, and −i[Hi,⋅]+DΩ-i[H_i, \cdot] + \mathcal{D}_\Omega generates trace-preserving CP maps, which do not increase the trace norm; so ∥etLi0Y∥1≤e−μit∥Y∥1\lVert e^{t\mathcal{L}_i^0} Y \rVert_1 \leq e^{-\mu_i t}\lVert Y \rVert_1, Li0\mathcal{L}_i^0 is invertible on traceless operators with ∥(Li0)−1∥1→1≤1/μi\lVert (\mathcal{L}_i^0)^{-1} \rVert_{1 \to 1} \leq 1/\mu_i, and it has exactly one stationary state ρlin\rho_{\mathrm{lin}}, the limit of its flow. If P(ρlin)≤2/7P(\rho_{\mathrm{lin}}) \leq 2/7, then Li[ρlin]=Li0[ρlin]=0\mathcal{L}_i[\rho_{\mathrm{lin}}] = \mathcal{L}_i^0[\rho_{\mathrm{lin}}] = 0: a stationary state that is not viable. If P(ρlin)>2/7P(\rho_{\mathrm{lin}}) > 2/7 and P(Γ)≤2/7P(\Gamma) \leq 2/7, put Y=Γ−ρlinY = \Gamma - \rho_{\mathrm{lin}}; then ∥Li[Γ]∥1=∥Li0Y∥1≥μi∥Y∥1\lVert \mathcal{L}_i[\Gamma] \rVert_1 = \lVert \mathcal{L}_i^0 Y \rVert_1 \geq \mu_i \lVert Y \rVert_1 and P(ρlin)−2/7≤P(ρlin)−P(Γ)=−2Tr(ρlinY)−Tr Y2≤2P(ρlin) ∥Y∥1P(\rho_{\mathrm{lin}}) - 2/7 \leq P(\rho_{\mathrm{lin}}) - P(\Gamma) = -2\mathrm{Tr}(\rho_{\mathrm{lin}} Y) - \mathrm{Tr}\,Y^2 \leq 2\sqrt{P(\rho_{\mathrm{lin}})}\,\lVert Y \rVert_1, which is the bound; in particular no stationary state has P≤2/7P \leq 2/7. For the sufficient condition: Li0[σi]=−i[Hi,σi]+DΩ[σi]\mathcal{L}_i^0[\sigma_i] = -i[H_i, \sigma_i] + \mathcal{D}_\Omega[\sigma_i], so ∥ρlin−σi∥1≤∥Li0[σi]∥1/μi\lVert \rho_{\mathrm{lin}} - \sigma_i \rVert_1 \leq \lVert \mathcal{L}_i^0[\sigma_i] \rVert_1/\mu_i, and the same purity inequality with σi\sigma_i in place of ρlin\rho_{\mathrm{lin}} gives P(ρlin)>2/7P(\rho_{\mathrm{lin}}) > 2/7.

(d) A stationary state exists: the composite flow maps the compact convex set of states into itself (frozen, its generator is of GKSL form), so each time-tt map has a fixed point (Brouwer), and a limit of such points as t→0t \to 0 is stationary. At a stationary state (b) gives ∥Li[Xi]∥1≤∣g∣ s(Hint)<εV(i)\lVert \mathcal{L}_i[X_i] \rVert_1 \leq |g|\,s(H_{\mathrm{int}}) < \varepsilon_V^{(i)}, and (c) excludes P(Xi)≤2/7P(X_i) \leq 2/7.

(e) Put Y=Xi−ρ∗Y = X_i - \rho_* and split L=A+R\mathcal{L} = \mathcal{A} + \mathcal{R} with A=−i[H,⋅]+DΩ+μ(σ Tr−id)\mathcal{A} = -i[H, \cdot] + \mathcal{D}_\Omega + \mu(\sigma\,\mathrm{Tr} - \mathrm{id}) and R\mathcal{R} the regenerative term. By (b), Y˙=AY+(R(Xi)−R(ρ∗))+u\dot Y = \mathcal{A}Y + (\mathcal{R}(X_i) - \mathcal{R}(\rho_*)) + u with ∥u∥1≤∣g∣ s(Hint)\lVert u \rVert_1 \leq |g|\,s(H_{\mathrm{int}}), and ∥etAY∥1≤e−μt∥Y∥1\lVert e^{t\mathcal{A}}Y \rVert_1 \leq e^{-\mu t}\lVert Y \rVert_1 as in (c). Duhamel's formula gives ∥Y(t)∥1≤e−μt∥Y(0)∥1+∫0te−μ(t−s)(LR∥Y(s)∥1+∣g∣ s(Hint)) ds\lVert Y(t) \rVert_1 \leq e^{-\mu t}\lVert Y(0) \rVert_1 + \int_0^t e^{-\mu(t-s)}\bigl(L_{\mathcal{R}}\lVert Y(s) \rVert_1 + |g|\,s(H_{\mathrm{int}})\bigr)\,ds, and Gronwall's inequality applied to eμt∥Y(t)∥1e^{\mu t}\lVert Y(t) \rVert_1 gives the bound. The aggregate is a mean of the marginals, and the trace norm is convex.

(f) This is the robustness of an exponentially stable equilibrium under a bounded non-vanishing perturbation (H. K. Khalil, Nonlinear Systems, 3rd ed., Prentice Hall 2002, §9.2), applied to the marginal equation (b), whose forcing is bounded by ε=∣g∣ s(Hint)\varepsilon = |g|\,s(H_{\mathrm{int}}) whatever the rest of the composite does. In detail: with J\mathcal{J} the Jacobian at ρ∗\rho_* on the 48-dimensional space of traceless Hermitian matrices, solve JTQ+QJ=−I\mathcal{J}^{\mathsf T} Q + Q\mathcal{J} = -I and put V(y)=⟨y,Qy⟩V(y) = \langle y, Q y \rangle. Near ρ∗\rho_* the field is C1C^1 (non-degeneracy keeps P(ρ∗)P(\rho_*) off the kinks of the gate), so on a ball ∥y∥≤r\lVert y \rVert \leq r one has V˙≤−12∥y∥2+2∥Q∥ ∥y∥ ε\dot V \leq -\tfrac12\lVert y \rVert^2 + 2\lVert Q \rVert\,\lVert y \rVert\,\varepsilon: a sublevel set Ωc={V≤c}\Omega_c = \{V \leq c\} inside the ball is forward invariant once ε\varepsilon is small, and every trajectory in it ends in ∥y∥≤4∥Q∥ ε\lVert y \rVert \leq 4\lVert Q \rVert\,\varepsilon (all norms on the finite-dimensional space are equivalent, which gives CC). The unperturbed flow carries the compact KK into Ωc/2\Omega_{c/2} by a common time TT (each point enters the open interior at some time, and by continuity so does a neighbourhood; finitely many neighbourhoods cover KK; Ωc/2\Omega_{c/2} is forward invariant). The field is Lipschitz on the compact state space with some constant Λ\Lambda, so up to time TT the perturbed marginal stays within εTeΛT\varepsilon T e^{\Lambda T} of the unperturbed one and is in Ωc\Omega_c at time TT for ε\varepsilon small.

(g) The canonical extension and a permutation-invariant HintH_{\mathrm{int}} commute with the permutations of the factors, so a permutation-invariant initial state stays invariant, all marginals coincide, and Mk(X)=X1\mathcal{M}_k(X) = X_1; (b) is the equation. For a local coupling the commutator with a product is local, the flow keeps products, and the partial trace of the coupling term is −ig[H1mf,X1]-ig[H_1^{\mathrm{mf}}, X_1]. ■\blacksquare

Numerical check (test_viability_passes_to_the_aggregate_only_at_weak_coupling, test_canonical_aggregation_is_unique_and_the_octonion_product_is_dead). (a): for k=2k = 2 and dimension 3 the linear conditions have full rank, 729 of 729 unknowns, and the solution equals the mean marginal to 10−1310^{-13}; without permutation invariance 324 free parameters remain. The embodied holon of Theorem 9.3 (μ=1\mu = 1, anchor of pure weight 0.80.8) has P(ρ∗)=0.3115P(\rho_*) = 0.3115 and P(ρlin)=0.3223P(\rho_{\mathrm{lin}}) = 0.3223, so εV=0.03225\varepsilon_V = 0.03225. Coupled to a copy through HintH_{\mathrm{int}} diagonal in a basis of maximally entangled vectors (spread s=1.8246s = 1.8246), (d) guarantees living parts for g≤0.01768g \leq 0.01768; the marginal identity (b) holds at the stationary state to 10−1510^{-15}. From a maximally entangled pure start and from a product start the marginals of the composite with a generic coupling end at distance 0.0643 g0.0643\,g from ρ∗\rho_* at g=0.01g = 0.01 and 0.020.02 (f). Under backbone dominance (κ=0.1\kappa = 0.1, μ=3.5\mu = 3.5, LR≤29κL_{\mathcal{R}} \leq 29\kappa) the bound (e) holds at every sampled time from a maximally entangled start; the measured distance is at most 4.3%4.3\% of it.

What this changes. Theorem 9.1 wanted "the composite is a holon" and Theorem 9.2 wanted "a consistent aggregation and weak coupling". Part (a) fixes the aggregation; parts (d)–(f) prove that the aggregate of weakly coupled viable holons is viable and within O(g)O(g) of a part's state, at every stationary state, along every trajectory from a compact part of the basin, and from every initial state with explicit constants under backbone dominance; part (g) says in what sense the aggregate is a holon. Two features limit what can be read from it. The canonical aggregate sees only the marginals, so it is blind to the correlations that Theorem 9.3 is about: it cannot certify anything the parts do not already have (collective consciousness needs a different aggregation, and the theory does not fix one). And the weak-coupling condition cannot be dropped:

Theorem 9.6 (Strong coupling kills every marginal aggregate; the octonion product kills every uncoupled pair) [T]​

Theorem 9.6 [T]

(a) Let {Φn}n=149\{\Phi_n\}_{n=1}^{49} be an orthonormal basis of C7⊗C7\mathbb{C}^7 \otimes \mathbb{C}^7 of maximally entangled vectors (for instance Φmn=7−1/2∑jωjn ∣j⟩∣j+m⟩\Phi_{mn} = 7^{-1/2}\sum_j \omega^{jn}\,|j\rangle|j + m\rangle, ω=e2πi/7\omega = e^{2\pi i/7}), and Hint=∑nEn∣Φn⟩⟨Φn∣H_{\mathrm{int}} = \sum_n E_n |\Phi_n\rangle\langle\Phi_n| with pairwise distinct EnE_n. For two holons of the form of Theorem 9.5, every stationary state X(g)X(g) has marginals Xi(g)=I/7+O(1/g)X_i(g) = I/7 + O(1/g). Hence P(Xi(g))→1/7P(X_i(g)) \to 1/7, and every aggregation that factors through the marginals — the canonical M2\mathcal{M}_2 among them — gives a dead aggregate at strong coupling, although each part alone is viable.

(b) Let V:C7⊗C7→C7V: \mathbb{C}^7 \otimes \mathbb{C}^7 \to \mathbb{C}^7, V(ei⊗ej)=ei×ejV(e_i \otimes e_j) = e_i \times e_j, be the octonion (Fano) product. Then VV†=6IVV^\dagger = 6I; W=V/6W = V/\sqrt6 is a co-isometry, Π=W†W\Pi = W^\dagger W a rank-7 projection inside the antisymmetric subspace, and E×(X)=WXW†+Tr((I−Π)X) I/7\mathcal{E}_\times(X) = WXW^\dagger + \mathrm{Tr}((I - \Pi)X)\,I/7 is a G2G_2-covariant CPTP map. It is not consistent, and it kills uncoupled parts: P(E×(X))≤5/21<2/7P(\mathcal{E}_\times(X)) \leq 5/21 < 2/7 for every separable XX, and P(E×(σ⊗σ))≤1/7+67(1−P(σ)2)2<0.2523P(\mathcal{E}_\times(\sigma \otimes \sigma)) \leq 1/7 + \tfrac67\bigl(\tfrac{1 - P(\sigma)}{2}\bigr)^2 < 0.2523 for every viable σ\sigma.

Proof. (a) Write the composite generator as L0+gB\mathcal{L}_0 + g\mathcal{B} with B=−i[Hint,⋅]\mathcal{B} = -i[H_{\mathrm{int}}, \cdot]. L0\mathcal{L}_0 is continuous on the compact set of states, so bounded there by some cc, and at a stationary state ∥BX∥=∥L0[X]∥/g≤c/g\lVert \mathcal{B}X \rVert = \lVert \mathcal{L}_0[X] \rVert/g \leq c/g. B\mathcal{B} is anti-Hermitian for the Hilbert–Schmidt product; its kernel is spanned by the ∣Φn⟩⟨Φn∣|\Phi_n\rangle\langle\Phi_n|, and on the orthogonal complement (the off-diagonal elements in the Φ\Phi basis) it multiplies by −i(En−Em)-i(E_n - E_m), so ∥BY∥2≥min⁡n≠m∣En−Em∣ ∥Y∥2\lVert \mathcal{B}Y \rVert_2 \geq \min_{n \neq m}|E_n - E_m|\,\lVert Y \rVert_2 there. Hence X=∑npn∣Φn⟩⟨Φn∣+O(1/g)X = \sum_n p_n |\Phi_n\rangle\langle\Phi_n| + O(1/g), and Tr2∣Φn⟩⟨Φn∣=Tr1∣Φn⟩⟨Φn∣=I/7\mathrm{Tr}_2|\Phi_n\rangle\langle\Phi_n| = \mathrm{Tr}_1|\Phi_n\rangle\langle\Phi_n| = I/7 for a maximally entangled vector. (b) VV†=6IVV^\dagger = 6I because each index kk lies on three Fano lines, each giving two ordered pairs; VV is antisymmetric, VS=−VVS = -V for the swap SS, so Π≤(I−S)/2\Pi \leq (I - S)/2; and V(ga⊗gb)=gV(a⊗b)V(ga \otimes gb) = gV(a \otimes b) for g∈G2g \in G_2. Put w=Tr(ΠX)w = \mathrm{Tr}(\Pi X): then E×(X)=A+(1−w)I/7\mathcal{E}_\times(X) = A + (1 - w)I/7 with A≥0A \geq 0, TrA=w\mathrm{Tr}A = w, and P=TrA2+2w(1−w)/7+(1−w)2/7≤1/7+6w2/7P = \mathrm{Tr}A^2 + 2w(1 - w)/7 + (1 - w)^2/7 \leq 1/7 + 6w^2/7. For X=σ⊗σX = \sigma \otimes \sigma, w≤Tr(I−S2σ⊗σ)=(1−P(σ))/2<5/14w \leq \mathrm{Tr}\bigl(\tfrac{I - S}{2}\sigma \otimes \sigma\bigr) = (1 - P(\sigma))/2 < 5/14 when P(σ)>2/7P(\sigma) > 2/7. For a pure product, w=∥a×b∥2/6w = \lVert a \times b \rVert^2/6; with a=x+iya = x + iy (x,yx, y real) the map b↦a×bb \mapsto a \times b has operator norm at most ∣x∣+∣y∣≤2 ∥a∥|x| + |y| \leq \sqrt2\,\lVert a \rVert, because b↦x×bb \mapsto x \times b has norm ∣x∣|x|; so w≤1/3w \leq 1/3, by convexity for every separable XX, and P≤1/7+6/63=5/21P \leq 1/7 + 6/63 = 5/21. ■\blacksquare

Numerical check. Two copies of the holon above, coupled through the Bell-basis HintH_{\mathrm{int}} (energies uniform in [−1,1][-1, 1]): the stationary marginals have P=0.3115P = 0.3115 at g=0.0177g = 0.0177 (the guaranteed threshold), 0.31140.3114 at 0.10.1, 0.31020.3102 at 0.30.3, 0.29700.2970 at 11 — still viable far beyond the threshold, which is conservative — then 0.22270.2227 at 33 and 0.15590.1559 at 1010 (canonical aggregate 0.15480.1548), while the purity of the joint state on C49\mathbb{C}^{49} stays at 0.030.03–0.100.10. For (b): the bound ∥a×b∥2≤2\lVert a \times b \rVert^2 \leq 2 is attained at a=(e1+ie2)/2a = (e_1 + ie_2)/\sqrt2, b=(e3−ie6)/2b = (e_3 - ie_6)/\sqrt2; over random pure products the aggregate never exceeds P=0.207P = 0.207, and over identical viable pairs 0.1470.147.

Routes that fail. The G2G_2-covariant octonion product (b) is the aggregation the Fano structure suggests, and it is dead on every uncoupled pair. The Petz recovery map of the partial trace, with reference σ⊗σ\sigma \otimes \sigma, runs the other way, from D(C7)\mathcal{D}(\mathbb{C}^7) to D(C49)\mathcal{D}(\mathbb{C}^{49}), and supplies no aggregation. The self-model of the composite acts on C49\mathbb{C}^{49} and does not reduce the dimension. An exact (HOL) — an autonomous generator on D(C7)\mathcal{D}(\mathbb{C}^7) satisfying A1–A5 that the aggregate follows — is not available in general: the forcing in (g) depends on the correlations, which the aggregate does not see; it is exact for local couplings and holds up to ∣g∣ s(Hint)|g|\,s(H_{\mathrm{int}}) in general.


We have travelled from the existence of dynamics through self-reference and No-Zombie to emergence. Now let us turn to another key block: how to check whether a system is alive? It turns out all viability conditions can be reduced to a single elegant criterion.

Unified Viability Condition​

The seven raw stress scores define the panel region Vσ={Γ:∥σsys(Γ)∥∞<1}V_\sigma=\{\Gamma:\|\sigma_{\mathrm{sys}}(\Gamma)\|_\infty<1\}. A component reaching one identifies the threshold crossed in this panel. Biological health would require a separately calibrated observation and outcome.

Diagnostic panel scope​

The seven raw scores define a separate region V_sigma; their U-condition implies structural majority, but they neither equal the four-part Cap2 gate nor imply biological viability. Uniform-diagonal Cap2 states already fail the raw L-condition. Use the exact four margins in the definitions.

Definitions and exact margins | Reconstruction identifiability

Sensorimotor Encoding​

Every living organism exists in the cycle "perception — decision — action — evaluation". A bacterium senses a sugar gradient, swims towards it, obtains nutrition — or not, and corrects its course. A human sees danger, chooses a path, evaluates the result. CC formalises this cycle precisely, with no free parameters.

Theorems 11.1–11.4 describe four facets of the sensorimotor cycle: encoding of the environment (how the world enters the system), optimal action (how the system responds), completeness of description (why three channels suffice), and hedonic valence (how the system evaluates whether it is "good" or "bad").

Observation-to-control interfaces​

T-100 specifies an observation category and a channel or control interface. Verify positivity and any composition laws on that interface. Distinct encoders can satisfy the same covariance equation. Compare them through the observation-law fibres and external calibration.

Definitions and exact margins | Reconstruction identifiability

Theorem 11.2 / T-101 (Optimal Action) [T]​

In Plain Terms

How does the system decide what to do? The answer is elegant: minimise the maximum stress. Recall the instrument-panel analogy from Theorem 10.1. The optimal action is one that leads to a state where none of the gauges is "in the red" — or, if they are all in the yellow, then with the least critical one.

This is a minimax strategy: instead of optimising a single metric (as in RL — reward), the system optimises the worst of seven indicators. This ensures robustness: the system does not sacrifice logic for dynamics, and does not sacrifice integration for articulation.

For an AI engineer: this is a ready-made utility function for an agent — without the need to engineer a reward.

Connection: Stress tensor, Motor stress

Statement [T]

The optimal action of a holon is determined by minimising the sup-norm of the stress tensor:

a∗=arg⁡min⁡a∈A∥σsys(Γ(τ+δτ∣a))∥∞a^* = \arg\min_{a \in \mathcal{A}} \|\sigma_{\mathrm{sys}}(\Gamma(\tau + \delta\tau \mid a))\|_\infty

where Γ(τ+δτ∣a)\Gamma(\tau + \delta\tau \mid a) is the predicted state under action aa.

Design criterion. Choose calibrated losses on a specified action set. A finite nonempty set has at least one minimizer. Uniqueness requires a single minimizing action or a supplied tie rule.

See: Sensorimotor theory

Target-dependent motor scores​

Define sigma_k=1-rho_kk/target_kk only for positive target populations. With a frozen target its derivative is -1/target_kk. If the target depends on rho, differentiate the target too. Zero total stationary drift does not imply zero regeneration or equality to the anchor. General G2 rotations do not preserve diagonal ratios.

Definitions and exact margins | Reconstruction identifiability

Admissible additional generators​

Nonnegative sums of GKSL generators are admissible; a signed difference need not be. A replacement term can itself be GKSL, so Hamiltonian/dissipation/regeneration is a grouping convention. The thermodynamic work/heat/matter dictionary is conditional on reservoir and charge choices, not a universal theorem of quantum channels.

Definitions and exact margins | Reconstruction identifiability

Theorem 11.4 / T-103 (Hedonic Valence) [T] + [I]​

In Plain Terms

How does the system know whether it feels "good" or "bad"? By the change in purity. If purity is growing — the system is "recovering", and this is experienced as positive valence (pleasure, satisfaction). If it is falling — as negative (pain, discomfort).

The formula Vhed\mathcal{V}_{\text{hed}} is not an abstract measure: it is the derivative of purity with respect to the regenerative channel. That is: "how fast am I recovering right now?" For a runner: the feeling "I chose the right pace" is positive Vhed\mathcal{V}_{\text{hed}}. The feeling "I am overloaded" is negative.

Important: the formula is a theorem [T], but the interpretation of it as a subjective experience is [I]. Mathematics says what the derivative equals. Philosophy says how it is experienced.

Connection: Purity dynamics, Replacement channel, Interiority

Statement

Hedonic valence is defined by the derivative of purity with respect to the regenerative channel:

Vhed:=dPdτ∣R=2κ(Γ)⋅gV(P)⋅Tr(Γ⋅(ρ∗−Γ))\mathcal{V}_{\text{hed}} := \left.\frac{dP}{d\tau}\right|_{\mathcal{R}} = 2\kappa(\Gamma) \cdot g_V(P) \cdot \mathrm{Tr}(\Gamma \cdot (\rho_* - \Gamma))

Epistemic stratification:

  • Formula — [T]: identity from the evolution equation
  • Observability at L2 (R≥1/3R \geq 1/3) — [T]: from T-77 (the replacement channel provides access to dP/dτdP/d\tau)
  • Phenomenal interpretation (connection with experience) — [I]

Proof. From the evolution equation: dP/dτ=−2Tr(Γ⋅DΩ[Γ])+2Tr(Γ⋅R[Γ,E])dP/d\tau = -2\mathrm{Tr}(\Gamma \cdot \mathcal{D}_\Omega[\Gamma]) + 2\mathrm{Tr}(\Gamma \cdot \mathcal{R}[\Gamma, E]). The Hamiltonian term does not change PP. Substituting R=κ(Γ)(ρ∗−Γ)⋅gV(P)\mathcal{R} = \kappa(\Gamma)(\rho_* - \Gamma) \cdot g_V(P) gives the formula. ■\blacksquare

See: Sensorimotor theory


The sensorimotor cycle is described. Now let us turn to attractors — equilibrium states toward which the system strives. These theorems, proved in core/dynamics, play a key role in CC, because the attractor is the system's "target self": the state it "wants" to reach.

Attractor and Structure Theorems​

Canonical definitions

The following theorems are proved in the core documentation and play a central role in CC. Here is a brief summary with cybernetic interpretation.

TheoremEssenceRole in CCCanonical definition
T-96 [T]Attractor is non-trivial: P(ρΩ∗)>1/7P(\rho^*_\Omega) > 1/7Every coherent system has a target stateEvolution
T-98 [T]Balance formula P(ρ∗)P(\rho^*) via κ/λgap\kappa/\lambda_{\mathrm{gap}}Basis of attractor hierarchy, stability radiusEvolution
T-77 [T]Replacement channel Φrepl\Phi_{\mathrm{repl}} — mechanism of reflectionAt L2, makes T-103 hedonics observableLindblad operators
T-78 [T]φ\varphi as a CPTP channel with Kraus representationBridge from categorical self-model to physicsSelf-observation
T-62 [T]Physical realisation of φ\varphi through spectral decomposition of L0\mathcal{L}_0Constructive formula for φSelf-observation
T-93 [T]PG(2,2)≅H(7,4)\mathrm{PG}(2,2) \cong H(7,4) — isomorphismStructure of Gap spaceGap dynamics
T-94 [T]Exponential memory kernel from compactnessJustification of non-Markovian extensionGap dynamics
T-80 [T]Gap bounded by sum of sector parametersEstimate of inter-sector gapsBerry phase
T-85Reality of the Hermitian-product traceIts imaginary part vanishes. A Berry term needs an independently specified connection.
T-82 [T]Uniqueness of the Fano operatorCC has no alternatives among Γ ⁣oct\Gamma_{\!\text{oct}}-covariant (Fano-structured) theoriesLindblad operators

Conclusion: the Theorem Landscape​

Let us retrace the route we have taken — but now from a bird's-eye view.

Foundation (Theorems 6.x): Dynamics exists and is physically correct. This is the "zero check" — without it, the subsequent results would be meaningless.

Self-reference (Theorems 7.x): Viability requires self-modelling. A system that does not observe itself is doomed. Iterative reflection converges to the unique fixed point — a stable "self-image".

No-Zombie (Theorem 8.1 and corollaries): the exact purity balance gives a stationary E-floor only under explicit rate, source and required-coherence bounds. The philosophical interpretation requires an E-to-phenomenality bridge; no universal floor is proved.

Composition and correlation (Theorems 9.1–9.6). Weak coupling and an interior margin allow the canonical mean marginal to stay within the chosen viability region. Strong coupling can destroy this property. Correlation of the joint stationary state depends on the interaction and anchor.

Diagnostics (Theorem 10.1): All viability conditions are equivalent to one: ∥σsys∥∞<1\|\sigma_{\mathrm{sys}}\|_\infty < 1. The stress tensor is a universal monitoring tool.

Sensorimotor cycle (Theorems 11.x): The system perceives the world (Enc), acts optimally (minimax stress), experiences the result (hedonic valence). Three channels — all that is needed; a fourth does not exist.

Attractors and structure (T-96, T-98, T-77, T-82, etc.): Every system evolves toward a non-trivial equilibrium. The balance between dissipation and regeneration determines "health". The Fano structure is unique — CC has no alternatives. Full formulations and proofs — in the summary table.

The mathematical results use the five axioms together with each named assumption and regime; the conditional No-Zombie result specifically requires rate and source bounds, and its phenomenal interpretation requires an explicit bridge (fractal closure and scale invariance, Theorems 9.1–9.2, hold at weak coupling by Theorem 9.5 — their earlier assumptions (HOL) and (AGG) are needed only beyond it, where Theorem 9.6 shows the transfer can fail; emergence, Theorem 9.3, needs (ND) for its weak-coupling criterion, and Theorem 9.4 proves (ND) for almost every anchor).


Dependency Map​

How to read the diagram: an arrow A→BA \to B means "theorem AA is used in the proof of theorem BB". Colours: blue — fundamental results (L-unification, attractor), green — key structural theorems (completeness), yellow — applied corollaries (diagnostics, capacity).

See: Dependency hierarchy for the full structure Ω → χ_S → L_k → ℒ_Ω → φ


What We Have Learned​

Let us summarise. In this chapter we have traversed the full path from basic existence theorems to the deepest results about the nature of consciousness:

  1. Dynamics exists and is correct (Theorems 6.1–6.2 [T]): the evolution equation has a unique solution preserving the physical meaning of the matrix Γ\Gamma (Hermiticity, positivity, normalisation).

  2. Self-models. A purity threshold does not imply a self-model. To test a proposed self-model, specify its action, observables and control class (§7.1).

  3. Conditional E-requirement (Theorem 8.1): purity balance and explicit rate/source bounds determine whether a positive E-floor is required. Identification with phenomenal necessity remains a separate bridge.

  4. Composition works at weak coupling (Theorems 9.1–9.6): the canonical aggregate of weakly coupled viable embodied holons — the mean marginal, which is unique — is viable, and its purity, reflection, integration and Gap profile lie within O(g)O(g) of a part's (fractal closure and scale invariance, [T at weak coupling], Theorem 9.5); the threshold on the coupling is explicit, and it cannot be dropped — at strong coupling the aggregate of two viable holons can be I/7I/7 (Theorem 9.6).

  5. Seven-score panel (Theorem 10.1 / T-92). The sup-norm tests membership in VσV_\sigma. Cap2 uses its own four margins. Diagnostic panel scope

  6. The sensorimotor cycle is closed (Theorems 11.1–11.4 [T]): environmental encoding is unique (up to G2G_2-calibration), action is optimal (minimax stress), three channels exhaust all possibilities, hedonics = dP/dτ∣RdP/d\tau|_{\mathcal{R}}.

  7. Structure is unique (T-82 [T]): the Fano operator is unique — CC has no alternatives among Γ ⁣oct\Gamma_{\!\text{oct}}-covariant (Fano-structured) theories in 7 dimensions.

Bridge to the Next Chapter

We have proved the theorems — but about what do they speak? What is the subject domain of CC? Do other interpretations of the axioms exist, beyond 7×77 \times 7 density matrices? In the next chapter we will engage with the model theory of CC: define the formal signature (language of the theory), construct the standard model (canonical interpretation), investigate questions of soundness and completeness, and then build functor bridges to other theories of consciousness (IIT, FEP, GNW). This is the transition from "what has been proved?" to "what is all this about?" — and "how does it connect to the rest of science?"


Related Documents: