Formalization of the Self-Modeling Operator φ
This is the sole canonical definition of the self-modeling operator . All other documents must reference this page rather than repeat the definition.
In the ∞-categorical framework the operator φ is understood not as a single morphism, but as a representative of a class of homotopically equivalent morphisms:
-
Multiplicity of paths: In the ∞-topos the mapping space is contractible, but contains many paths (morphisms) connected by homotopies.
-
φ₀ as canonical representative: The concrete operator defined in this document is a representative of its homotopy equivalence class . The choice of is made by the minimality criterion — minimization of divergence from the self-model.
-
Freedom of choice: The existence of alternative representatives in the same class reflects the fundamental free will — a system can realize different paths to the same attractor.
-
Relation to Ω⁷: The choice of a concrete representative is consistent with the Ω⁷ axiom, where the seven-dimensional structure fixes the canonical basis for decomposition.
Summary table of definitions of φ
This document considers four equivalent definitions of the operator :
| Definition | Formula | Context | Status |
|---|---|---|---|
| Replacement channel | Canonical (T-62) | [T] | |
| CPTP via Kraus operators | General form | [T] | |
| Fano E-accentuation | preserving coherences | Theorem 8.1 (No-Zombie) | [T] |
| Categorical functor | ∞-topos | [T] |
The replacement channel is the canonical physical realization of the self-modeling operator (proof). Here is the categorical self-model of the current state [T], is the degree of self-modeling. The channel is exact at (full convergence to ), but for intermediate values of realizes approximate self-modeling — the system is in a dynamic balance between its current state and its internal model. The remaining three definitions are equivalent to the replacement channel via the equivalence theorem [T].
Stratification of definitions
The operator is defined through the stationary state , not the other way around:
All components of the chain have independent definitions: — via primitivity of the linear part [T-39a], — via the distance from to , parameter — via . There is no circularity: the full hierarchy of levels 0–9 is in the Ω⁷ axiom.
Categorical definition of φ
This section establishes an independent categorical definition of the operator via a universal property, eliminating any apparent circularity in the definitions.
φ as a left adjoint to the inclusion of subobjects
In the ∞-topos generated by the Ω⁷ axiom, the self-modeling operator is defined as the left adjoint functor to the inclusion of the category of subobjects:
where:
- — the category of logically consistent subobjects of (satisfying the internal logic )
- — the canonical inclusion (embedding)
- — adjunction: is left adjoint to
Universal property: For any object and any subobject :
Theorem on the equivalence of three definitions of φ
The three definitions of the operator φ are strictly equivalent:
Theorem (Equivalence of definitions of φ):
The following definitions specify the same operator :
| # | Definition | Formula | Source |
|---|---|---|---|
| 1 | Categorical | Left adjoint | |
| 2 | Dynamical | Limit of evolution | |
| 3 | Idempotent | , | Projection with fixed point |
Proof of equivalence:
(1) ⟹ (2): Categorical ⟹ Dynamical
- The left adjoint to the inclusion projects onto the invariant subspace
- annihilates : for
- By the Perron–Frobenius theorem for CPTP channels:
- The invariant projector by uniqueness of the left adjoint ∎
Step (1) ⟹ (2) uses the Perron–Frobenius theorem for the linear part . Primitivity of is proven for all viable holons: from (AP)+(PH)+(QG)+(V) the interaction graph is connected (otherwise the system decomposes into blocks with , contradicting the minimality theorem), and connectivity of + atomic operators give a trivial commutant by the Evans–Spohn criterion (Evans 1977, Spohn 1976), so .
Important (self-referential fix, T-96): the full generator is nonlinear and is not primitive — it has a nontrivial zero-mode . Definition (2) below is therefore the projector onto the multiplicity-1 zero mode of the linearised full generator at , not applied as a linear semigroup (which does not even make sense for nonlinear , and would collapse to ). The equivalence (1) ⟺ (2) ⟺ (3) [T] is read with this convention. Full proof: Primitivity of ℒ₀; attractor: T-96.
(2) ⟹ (3): Dynamical ⟹ Idempotent
- (idempotency)
- Fixed point: for any , then ∎
(3) ⟹ (1): Idempotent ⟹ Categorical
- An idempotent map with defines a reflector
- A reflector is automatically left adjoint to the inclusion
- Universal property: follows from idempotency ∎
Direction (1)⟹(2) follows from primitivity of the linear part [T]: the left adjoint projects onto the invariant subspace, and primitivity provides the spectral gap and convergence of the linear dynamics. Direction (2)⟹(1): any minimizer of the variational functional under the CPTP condition is a stationary point, and the CPTP contraction φ guarantees uniqueness . Thus (2)⟹(1) is also [T] via the categorical definition of φ. [✗] Retracted 2026-09-25: this argument rested on the variational characterisation of φ (Theorem 3.1 of the FEP derivation, registry row 39e), which is retracted: the functional equals , is linear in and is minimised by the projection onto the top eigenvector of , not by . The equivalence does not need it: the cycle (1)⟹(2)⟹(3)⟹(1) above is proved without the variational functional, and (2)⟹(1) is the composite (2)⟹(3)⟹(1). The three steps of the cycle keep status [T].
Independence from coherence levels
Critically: and the coherence levels are defined independently of each other, both constructions are derived from :
| Construction | Source | Definition |
|---|---|---|
| Levels | Stratification by the logical Liouvillian | |
| Operator | Left adjoint to the inclusion |
This eliminates any circularity: both notions are consequences of the structure , not defined through each other.
See Dependency hierarchy for the full diagram: Ω → χ_S → L_k → ℒ_Ω → φ.
φ(Γ) as best approximation
Interpretation: is the best approximation of the state in the category of logically consistent subobjects.
Formally, is the coreflector:
Geometric intuition: "projects" an arbitrary state onto the nearest logically consistent state — this is the categorical analogue of orthogonal projection onto a subspace.
Theorem: φ as stationary distribution
Theorem (φ as limit of logical evolution):
Let be the logical Liouvillian generated by the internal logic . Then:
Proof:
-
The logical Liouvillian generates a semigroup on .
-
The invariant objects of this semigroup are exactly the subobjects from :
-
By the convergence theorem for primitive CPTP channels (analogue of Perron–Frobenius for quantum channels) applied to the linear part , the projector onto its multiplicity-1 zero mode (evaluated at the linearisation about ) exists. Primitivity of for viable holons [T] — see proof. Here denotes this zero-mode projection, not the constant map onto (which is what naive primitivity of the full would give — the self-referential-ρ* fix, T-96).
-
This projection coincides with the coreflector by uniqueness of the left adjoint. ∎
Corollary: is the stationary distribution of the logical dynamics — the attractor of evolution under .
Strict Mathematical Theory
Contents
- Introduction and motivation
- Formal definition of φ
- Theorem on existence of fixed point
- Relation to reflection measure R
- Categorical aspect
- Corollaries and limitations
- Implementation requirements
- Operational algorithm for φ
- Relation to the regeneration mechanism
1. Introduction and motivation
1.1 The problem
In UHM, self-observation is defined via the conditions:
- contains a subsystem
- Reflexive closure:
However, the operator lacks a rigorous definition. This document fills that gap.
1.2 Requirements for formalization
The operator must satisfy:
- Mathematical correctness: — defined on the space of operators
- Structure preservation: is a density matrix if is a density matrix
- Physical interpretability: models the process of self-observation
- Existence of a fixed point: under certain conditions
2. Formal definition of φ
2.1 Preliminary definitions
Definition 2.1 (Space of density matrices):
For (seven-dimensional space of the Holon):
Definition 2.2 (Metric on ):
is a complete metric space (closed subset of ).
2.2 Definition via reduced density matrix
Definition 2.3 (Self-modeling operator — reduction form):
Let system with coherence matrix be decomposed into:
- Model subsystem
- Remaining system (model environment)
— model environment. Not to be confused with — the Interiority dimension.
Then:
where:
- — density matrix of the extended system
- — partial trace over the model environment
Problem: This definition requires an extended space and is not closed on .
2.3 Definition via predictive model (main definition)
Definition 2.4 (Self-modeling operator — predictive form):
Let the system possess an internal predictive model represented by a CPTP map:
where CPTP = Completely Positive Trace-Preserving.
— predictive CPTP map. Not to be confused with — the integration measure.
Self-modeling operator:
Constructive definition of :
is constructed via Kraus operators :
Interpretation of Kraus operators:
- — "perception filters" of the system
- Each corresponds to a partial aspect of self-observation
- The condition guarantees preservation of normalization
The operator φ does not violate the no-cloning theorem (Wootters–Zurek, 1982). The key distinction:
- No-cloning excludes the existence of a unitary operator such that for arbitrary . Cloning is exact unitary copying of an unknown state.
- Self-modeling φ is a CPTP channel (Kraus representation), not a unitary operation. CPTP channels are fundamentally irreversible: they decrease state distinguishability ( by fidelity monotonicity). The self-model is an approximate, coarse-grained projection, not an exact copy.
Formally: with guarantees — the purity of the self-model does not exceed the purity of the original. This is categorically different from cloning, where .
2.4 Parameterization via projections
Definition 2.5 (Projection self-modeling operator):
The most natural physically motivated form:
where:
- — orthogonal projectors,
- — "depth of self-observation"
- — prior model (may be or other)
With and :
For :
This is "dephasing self-observation" — preserves the diagonal in the basis .
2.5 Contracting self-modeling operator
Definition 2.6 (Contracting operator):
To ensure existence of a fixed point:
where:
- — contraction parameter
- — any CPTP map
- — fixed "anchor" point (e.g., maximally mixed state )
Lemma 2.1: is a contracting map with constant .
Proof:
(for a unital , which does not increase the Frobenius norm). ∎
Scope (corrected 2026-09-25). The lemma needs three things: constant, the anchor constant, and unital — or the trace norm in place of the Frobenius norm, which every CPTP map contracts. The parenthetical "CPTP does not increase the Frobenius norm" is false for non-unital channels: on stretches the Hilbert–Schmidt distance of and by (test_self_model_contraction_holds_only_for_constant_weight_and_unital_part). The UHM self-models have , which depends on , and they are not contractions: the Lipschitz constant of at a pure state is , that of at a basis state . Their fixed points exist anyway (Brouwer) and are unique by direct computation — for , for (evolution); contracts only toward its own fixed point, .
2.6 Canonical form of φ for UHM
This section defines the canonical construction of the self-modeling operator for UHM. This is a concrete specification linking the abstract definitions above to the seven-dimensional structure of the Holon.
Definition 2.7 (Canonical form of φ for UHM):
The canonical form of the self-modeling operator:
where:
- for small (typical value: )
- — predictive CPTP channel (defined below)
Definition 2.8 (Predictive CPTP channel):
with Kraus operators:
where are orthogonal projectors onto the basis states :
CPTP condition (verification):
The weights are realized NOT by modifying the Kraus operators, but via a weighted mixture of basic channels or by modifying the anchor state . See below.
Base predictive channel (dephasing in measurement basis):
This channel preserves the diagonal and destroys coherences.
Definition 2.9 (Weighted self-observation via anchor):
To model varying "depth of self-observation" across dimensions, a weighted anchor is used:
| Weight | Interpretation |
|---|---|
| Attention to distinctions (Articulation) | |
| Awareness of patterns (Structure) | |
| Perception of time flow (Dynamics) | |
| Logical reflection (Logic) | |
| Phenomenal self-awareness (Interiority) | |
| Connection to deep foundation (Foundation) | |
| Integration into unified Self (Unity) |
Special case: uniform self-observation
With for all :
— maximally mixed state.
Definition 2.10 (E-accentuated self-observation):
Systems with conscious experience are characterized by an accentuation of dimension :
At : the anchor approaches the pure state .
Theorem 2.1 (Fixed point of canonical φ):
For with there exists a unique fixed point:
Proof:
Since is a diagonal matrix:
Therefore:
Uniqueness follows from contractivity at (Banach theorem). ∎
Special case (uniform anchor):
With for all : — maximally mixed state.
Corollary 2.1: With a uniform anchor the fixed point is the maximally mixed state.
For a uniform anchor: .
The fixed point of uniform self-observation is NOT viable!
This means:
- Ideal uniform self-knowledge is incompatible with viability
- Living systems exist in a dynamic balance away from the fixed point
- Regeneration keeps the system in the region
Definition 2.11 (Viable anchor)
To ensure a viable fixed point, the anchor must satisfy:
The E-accentuated anchor is not an arbitrary choice, but a consequence of the L2-definition of consciousness.
Theorem 2.2 (E-accentuation from L2-definition):
Let the system satisfy the cognitive qualia condition (L2):
- — reflection
- — integration
Then its anchor state is necessarily E-accentuated:
Proof:
-
Consciousness measure [Т T-140] and the separate viability condition (differentiation by E).
-
Reduced matrix singles out the Interiority dimension as privileged.
-
For L2-systems: High requires a rich structure precisely in .
-
Consequence for anchor: The self-model of a conscious system inevitably accentuates E — the dimension through which the system is aware of itself.
-
Formally: Minimization of subject to gives:
Solution: at . ∎
Corollary 2.2: The uniform anchor () corresponds to systems without self-awareness (L0/L1), for which the question of viability of the fixed point does not arise — they do not strive toward φ(Γ).
Canonical value of α:
For systems at the L2 boundary (, ):
Example: E-accentuated anchor with (conservative estimate):
has . ✓
E-accentuation is not a "privilege" of dimension E, but a structural consequence of the fact that conscious systems are defined through experience. Non-conscious systems (L0) do not have this constraint — their anchor can be uniform, and the question is not relevant for them (see theorem on critical purity).
The choice of anchor depends on the interiority level (L2), which is defined via R, which is defined via φ. The circularity is only apparent: the canonical and the self-registering are closed-form functions of the current (φ operator), with read on the same , so no fixed point has to be found first. The convergence theorem below settles the second question — whether iterating "self-model → attractor → self-model" converges — and it does so only for an embodied holon under backbone dominance. (Until 2026-09-25 this box said the circularity is resolved by T-191 for every holon.)
Theorem T-191 (Convergence of the φ-tower; restated 2026-09-25) [T]
The former statement — the tower converges for every holon, from any anchor, with rate — is retracted with its proof, which is the retracted proof of T-124c in another dress. Step 1 called a generator of a contractive CPTP semigroup with a unique stationary state; the scalars depend on , and the gate makes the flow bistable. For an isolated holon with a fixed target and the flow from stays at (the gate is closed) while the flow from ends at a living state with , so is not defined. Step 3 used , "verified in T-96" — T-96 verifies no such bound. For the canonical unital the tower started at stays at the dead replacement channel (dead isolation).
Let an embodied holon carry the backbone term (T-148); let be a trace-norm Lipschitz constant on of , uniform in the target state , and . If then the tower of self-models — any state, the stationary state of the dynamics with regeneration target (so that is the replacement channel ) — is well defined, and with one limit for every initial anchor: is the stationary state of the dynamics that regenerates toward itself.
Proof. Each iterate is defined. is the backbone dominance of T-124c (3): for every target the dynamics has exactly one stationary state , and every trajectory reaches it at rate in trace norm.
The iteration contracts. For targets the two generators differ by , whose trace norm is at most . Run the dynamics with target from ; the difference from the constant solution of the dynamics with target obeys, by the contraction estimate of T-124c (3) with this inhomogeneity, . As the trajectory reaches , so .
Banach. with the trace norm is complete; is a contraction with constant , so it has one fixed point , the iterates converge to it geometrically, and two towers started at approach each other as .
Constants. constant: and give . Witness (test_phi_tower_converges_only_under_backbone_dominance): , , , so ; from , and a random pure anchor the towers meet to , each step contracts by at most , and the limit has residual with the gate open; the isolated counterexample of the box above is in the same check.
Corollary (SAD tower). The Self-Awareness Depth tower (T-142 [T]) corresponds to the first three iterates. Under the hypothesis of the theorem the differences decrease geometrically. By T-142 [T], SAD — the fourth iterate would require , violating — so the tower terminates at finite depth whatever the hypothesis; convergence matters only for the realisable levels.
Dependencies: T-124c (3) [T] (backbone dominance: existence, uniqueness and rate of the stationary state), T-148 [T] (backbone term), T-142 [T] (SAD). Standard mathematics: Banach fixed-point theorem, the variation-of-constants estimate for a flow contracting in trace norm. (The dependencies read "T-39a (spectral gap), T-59, T-96 (), T-124c (attractor uniqueness)" until 2026-09-25; the uniqueness statement of T-124c is retracted, and T-96 bounds no .)
2.7 Spectral formula for φ (explicit computation)
This section provides an explicit computable formula for the operator via the spectral decomposition of the logical Liouvillian . This makes the theory fully constructive.
Theorem 2.3 (Spectral formula for φ):
where:
- — right and left eigenvectors of
- — eigenvalues of
- Sum over with (stationary modes)
- — inner product in vectorized space
Proof:
-
By definition (see Theorem: φ as stationary distribution):
-
Decomposition into eigenfunctions:
-
As :
- : (decay)
- : excluded by CPTP structure (divergence impossible)
- : bounded (stationary modes)
-
Therefore:
Primitivity of the linear part ensures a spectral gap. In the vicinity of the non-trivial attractor the formula simplifies to projection onto the zero mode (, multiplicity 1):
where is the stationary state of the full dynamics (categorical self-model, Definition 1), is the corresponding left eigenvector.
Algorithm for computing φ (spectral method):
mount core.math.linalg.{StaticMatrix, StaticVector, eig, inverse};
/// Compute φ(Γ) via spectral decomposition of the logical Liouvillian.
///
/// The Liouvillian ℒ_Ω is vectorised as a 49×49 superoperator; φ projects Γ
/// onto the kernel (stationary modes with Re(λ) ≈ 0).
public pure fn compute_phi_spectral(
gamma: &StaticMatrix,
l_omega: &StaticMatrix,
) -> StaticMatrix<Complex, 7, 7>
{
let (eigvals, r_vectors) = eig(l_omega);
let l_vectors = inverse(&r_vectors).unwrap().transpose(); // left eigenvectors
let gamma_vec = gamma.flatten(); // 49-vector
let mut phi_vec = StaticVector<Complex, 49>.zeros();
const TOL: Float = 1.0e-10;
for k in 0..49 {
if eigvals[k].real().abs() < TOL { // stationary mode
let coeff = l_vectors.column(k).conjugate().dot(&gamma_vec);
phi_vec = &phi_vec + r_vectors.column(k) * coeff;
}
}
let phi_gamma = phi_vec.reshape<7, 7>();
let hermitised = (&phi_gamma + phi_gamma.adjoint()) / Complex.from_real(2.0);
&hermitised / hermitised.trace() // renormalise Tr = 1
}
Computational complexity:
| Operation | Complexity |
|---|---|
| Spectral decomposition of | for , i.e. |
| Projection onto stationary modes | |
| Total complexity | , but is computed once |
Relation to contracting form:
The spectral formula is equivalent to the canonical definition with the correct choice of . Advantages of the spectral form:
- Explicit computation — no iterations required
- Uniqueness — no dependence on initial state
- Categorical consistency — corresponds to the left adjoint to inclusion
2.8 n-th order reflection (for L3/L4)
Defining levels L3 and L4 of the interiority hierarchy requires an iterated operator φ.
Definition 2.12 (Iterated operator φ):
with .
Definition 2.13 (n-th order reflection):
where — fidelity.
Thresholds for L3/L4:
| Transition | Threshold | Universal formula |
|---|---|---|
| L1→L2 | ||
| L2→L3 | ||
| L3→L4 | — |
Algorithm for computing :
mount core.math.linalg.matrix_sqrt;
/// Second-order reflection R^(2) = Fid(φ(Γ), φ(φ(Γ))).
/// Fidelity F(ρ₁, ρ₂) = |Tr √(√ρ₁ ρ₂ √ρ₁)|².
public pure fn compute_r2(
gamma: &StaticMatrix,
l_omega: &StaticMatrix,
) -> Float { 0.0 <= self && self <= 1.0 }
{
let phi_gamma = compute_phi_spectral(gamma, l_omega);
let phi_phi_gamma = compute_phi_spectral(&phi_gamma, l_omega);
let sqrt_phi = matrix_sqrt(&phi_gamma);
let inner = sqrt_phi.matmul(&phi_phi_gamma).matmul(&sqrt_phi);
let trace_sqrt = matrix_sqrt(&inner).trace().abs();
(trace_sqrt * trace_sqrt).clamp(0.0, 1.0)
}
3. Theorem on existence of fixed point
3.1 Main theorem
Theorem 3.1 (Existence of reflexion fixed point):
Let be a contracting map with constant :
Then:
and for any :
with convergence rate:
Proof:
Step 1: Completeness of the space
is a closed subset of the Banach space .
Checking closedness:
- The limit of a sequence of Hermitian matrices is Hermitian
- The limit of a sequence of positive semi-definite matrices is positive semi-definite (closed cone)
- is a continuous function,
Therefore, is a complete metric space.
Step 2: Applying the Banach theorem
is a contracting map on a complete metric space.
By the Banach fixed point theorem:
- There exists a unique fixed point
- Iterations converge to for any initial condition
Step 3: Structure preservation
Show that :
(by construction of or as CPTP map).
, where and for all .
is closed . ∎
Scope (2026-09-25). Theorem 3.1 is Banach's theorem and is true as a conditional; its hypothesis — a global contraction with constant — is met by the form of Lemma 2.1 with constant and anchor and a unital , not by the UHM self-models , , , whose weight varies with the state (Lipschitz constants and above; has at least eight fixed points). For them a fixed point exists by Brouwer's theorem on the compact convex , and uniqueness, where it holds, comes from the explicit computation.
3.2 Approximate fixed points
Definition 3.1 (-fixed point):
is called an -fixed point if .
Theorem 3.2 (Existence of -fixed point for non-contracting ):
Let be a continuous map (not necessarily contracting).
Then for any there exists such that:
Proof:
Consider the family of maps:
where is the center of .
For : if is non-expansive, is a contracting map with constant (as in Lemma 2.1). Corrected 2026-09-25: for a merely continuous this step fails, and the theorem needs no approximation at all — Brouwer's theorem gives an exact fixed point of every continuous on the compact convex , so is attained. The argument below is kept for the non-expansive case.
By Theorem 3.1: .
Consider:
where (diameter of the density matrix space).
Choosing , we get:
∎
3.3 Contraction conditions for CPTP maps
Theorem 3.3 (Contraction criterion):
A CPTP map is contracting with constant if and only if:
where is the restriction of to the orthogonal complement of .
Interpretation: is contracting if it has a unique invariant state and all perturbations decay.
Corrected 2026-09-25 [✗ as "if and only if"]. The spectral condition makes some power a contraction (and one in a suitably chosen norm); it does not make itself a contraction in the Frobenius norm. The channel on has the unique invariant state and its other eigenvalues in , yet stretches a Hilbert–Schmidt distance by . The converse (a contraction has a unique fixed point) is Banach.
Examples of contracting CPTP:
- Thermalization:
- Depolarizing channel:
- Amplitude damping:
Contracting for .
4. Relation to reflection measure R
4.1 Definition of R
Definition 4.1 (Reflection measure):
Equivalent form: , where , (square root of purity).
is the canonical definition used in all thresholds (). It is a measure of proximity to the maximally mixed state , NOT a measure of quality of self-modeling.
The quality of self-modeling is defined separately (formerly also written ; unified notation — the three working forms of R):
Comparison at characteristic states:
- At (dissipative attractor): , .
- At a pure state (): , depends on .
Sections 4.2–4.3 below analyse the convergence of the self-model quality and are written in ; in all other sections and in threshold conditions .
4.2 Convergence of R_φ as fixed point is approached
Theorem 4.1 ( as ):
Let be a contracting map with fixed point .
Then:
Proof:
As :
Therefore:
(The denominator is bounded away from zero for any density matrix: .) ∎
4.3 Estimate of rate of convergence of R_φ
Theorem 4.2 (Rate of convergence of ):
For contracting with constant and sequence :
where .
Proof:
For : , giving the bound:
∎
Primitivity of guarantees exponential convergence for any initial state , without additional conditions on initial data.
4.4 Relation of R to consciousness measure C
Theorem 4.3 (reflection factors at the fixed point):
The consciousness measure [Т T-140] takes the canonical ; the convergence results of §4.2–4.3 concern the self-model quality . At the fixed point the two factor cleanly:
with whenever lies inside the conscious window. (An earlier reading substituted into , yielding ; under the canonical that value is attained only at , where — the reading is withdrawn.)
Differentiation enters as a separate viability condition, not as a factor of .
Corollary: Ideal self-knowledge () maximizes the self-model-quality factor — the quantity carried by the meaning functional and the -towers — while the canonical factor of is pinned by purity.
4.5 The derivative of the self-model: Dφ
Everything dynamical about is carried by one object: how the self-model responds to a change of state. The exact flow identity of the three working forms of R,
contains the derivative of the self-model map — the response kernel of self-modelling. This subsection computes it for the canonical family and verifies the identity by an independent route.
Definition () [D]. The Gateaux derivative of at along a tangent direction (Hermitian, traceless): . For a Lipschitz (contracting) it exists almost everywhere (Rademacher, finite dimension); for the smooth families below — everywhere.
For the canonical dissipative family with :
preserves the Hermitian-traceless tangent space and is -equivariant.
Proof. Product rule on : the -slot contributes ; the -slot contributes with (since ); the reference is constant. Trace preservation: . Equivariance: , , are unitarily invariant and is the unique -invariant state, so commutes with conjugation by .
Two-route consistency (part of T-249) [T]. For this family the closed form holds pointwise, whence directly . Substituting Theorem 4.4 into the flow identity must give the same. Indeed, with :
(the latter by trace preservation, ), and . Collecting terms with :
Machine verification. Finite-difference check of and three-route agreement (identity / closed form / numerical derivative along random density-matrix paths): maximal discrepancy over random states.
4.6 The bandwidth theorem for R_φ
Along any trajectory of the dynamics,
For the canonical family, (numerically across the conscious window).
Proof. Cauchy–Schwarz on the flow identity, with by the definition of . For the family bound: and , , so the second term is .
Corollary (path-length law) [T]. Let — the mismatch amplitude. On segments with :
— reorganizing the self-model is paid for in state-space path length. Proof: , and the theorem's right side divided by collapses to .
Remark (discrete instance). Theorem 4.2 above is exactly this law along the -tower: , with the geometric path (T-191 convergence). The continuous and discrete forms are two readings of one bandwidth constraint.
Remark (). , , , are -invariant; the law is observer-independent.
Numerical anchor [I]. At the psychedelic-peak profile of altered states (; ): , so the trajectory must traverse at least Frobenius units of state motion between onset and peak. The collapse of the "I" costs actual movement of the state; a short (micro-dose) trajectory cannot produce it.
4.7 Mechanisms: dissolution, training, two timescales
Ego dissolution [C]. Fast state motion with a lagging self-model — large while stays comparable to — drives up at the bandwidth-permitted rate: the phenomenological "φ does not have time to restructure" becomes the quantitative statement that grows no faster, and generically as fast, as .
Two-timescale training model [С structure; [I] numbers]. Let the self-model carry a trainable target: , with evolving on the slow timescale of -restructuring (procedural memory; the T-155 learning channel). Then , and training that aligns with the practiced state raises the baseline at fixed — the mechanism behind the cumulative shift in the shamatha progression. The samādhi signature ( and rising simultaneously) is thereby resolved: the information the categorical carries beyond purity is exactly the learned target .
Training law (exact within the alignment model) [C]. If the slow variable follows the gradient-alignment flow toward a practiced state (the gradient of the alignment loss at mobility ), the solution is the convex path — automatically a density matrix for every — and the baseline self-model quality obeys the exponential saturation law
Practice approaches its ceiling exponentially, at twice the flow rate (: the distance enters squared), with the rate constant set by the slow T-155 channel — the cumulative shift of the shamatha progression acquires a closed form. Status [C] at the gradient-alignment model; within the model the law is exact (machine-checked against direct integration of the flow, steps).
The gate condition and the threshold. Define (G) [D]: the channel-class discriminator factors through the self-model readout, with classification accuracy bounded below by . Statement [C at (G)]: (G) implies — Bayesian plurality among three equiprobable alternatives requires accuracy above .
Register of the module (closed). Both items of the original register are closed below: (i) the gate is now a theorem-level bound — §4.9 (T-252); the residual freedom is the accuracy model and the exact placement of the working threshold inside the derived band [C]; (ii) for implicitly defined — §4.8 (T-251); the residual condition is -smoothness of the abstract categorical generator in the Bures topology [C], satisfied by every corpus-realized family.
4.8 Dφ for implicitly defined self-models
Theorem 4.4 covers the explicit canonical family. The categorical , however, is given implicitly — as the fixed point of a generator whose iteration is the T-191 tower. Its derivative follows from the implicit function theorem with a Neumann series — which is also the categorical answer: differentiating a reflector along its universal property is inverting minus the derivative in the model slot.
Let be with (uniform contraction in the model slot), and let be the unique fixed point of . Then is and, evaluated at ,
with ; consequently , and the bandwidth theorem T-250 applies to every contraction-defined self-model.
Proof. Set ; then is invertible by the Neumann series (). The finite-dimensional implicit function theorem yields with ; the series and the norm bound are the Neumann expansion.
Three readings. Tower: the -th term is the sensitivity transmitted through storeys of the T-191 tower — the geometric decay of storey-sensitivities is the tower's convergence, differentiated. Degenerate check: for independent of () the series collapses to — Theorem 4.4 (T-249) is the zeroth-order case. Machine verification: for a genuinely nonlinear generator (–), the predicted and the brute-force finite-difference derivative of the re-solved fixed point agree at (numeric Jacobians); the degenerate case reproduces T-249 at .
4.9 The gate theorem: discrimination through the self-model
This closes the gate condition (G) of §4.7. First the structure [Т — structural reading]: in the only feedback channel computed from the self-model is the regeneration — the system's sole endogenous corrective action reads the state through . Endogenous adaptive discrimination — discrimination the system can act on — is therefore -mediated by construction (the identification of "adaptive" with "-actionable" is definitional [D]).
Let be any POVM implementing a -hypothesis decision, and (traceless). Then:
(a) for every , and the outcome distributions satisfy ;
(b) , and the constant is tight on traceless ;
(c) hence the -mediated success probability obeys , where is the true-state accuracy of the decision rule, and Bayesian dominance is guaranteed whenever
Both inequalities of the chain are individually saturated: (a) by the Jordan projector of , (b) by the -split spectrum — so no smaller constants exist.
Proof. (a) Jordan-decompose ; tracelessness gives ; with : , and . Equality at (the positive-eigenspace projector): . (b) Maximize over , in dimension 7: the KKT condition forces at most one positive value ( copies) and one negative ( copies), ; then , increasing in the support size and maximal at the balanced split : . Witness attaining it exactly: (, , ratio ). Finally by the definition of . (c) Substitution.
Remark (the constant is an odd-dimension effect). The generic -dimensional constant is : in even dimension it equals exactly, in odd dimension it sits strictly below. At the balanced split is forced to be and gives . The naive rank bound (Cauchy–Schwarz) is therefore not attainable: tracelessness forbids the aligned spectra that would saturate it. (No structural weight is placed on the split here — it is the arithmetic of odd , recorded per the anti-numerology register.)
Corollary (the working threshold) [С — canonical alignment]. For with an ideal true-state discriminator () the sufficient bound sweeps the band across the conscious window ( at the viability edge, at the ceiling); the working threshold lies inside this derived band and is fixed at the canonical value by alignment with Char-R-III. What §4.7 posited as the bare condition (G) is now the theorem-level bound (a)–(c) with tight constants; the only remaining freedom is the accuracy model and the placement of the working value inside the band.
Corollary (sectoral gate: the per-channel threshold) [T]. Fix a coherence channel and its canonical three-outcome readout , , where projects onto and (a valid POVM: , ). The -mediated shift of each outcome probability is bounded by , and the off-diagonal part is exactly the sectoral reflection: by the definition of . In the coherence-dominated regime (diagonal mismatch negligible — precisely the regime in which the sectoral form is deployed) the per-channel discrimination loss is governed by alone, so the dominance argument transfers channel-wise verbatim: the sectoral working threshold carries the same theorem-level gate as the global . This derives the per-channel threshold previously inherited by analogy in the unconscious.
Machine verification. random -outcome POVMs: (a) and (b) hold with margin; Jordan-projector saturation of (a) at ; sharp constant confirmed — random traceless search reaches , the -witness attains it exactly; band endpoints and ; sectoral identity, POVM validity and shift bound on random states.
5. Categorical aspect
The categorical formalism provides additional structure for understanding , but is not necessary for practical computations in UHM. See also categorical formalism.
5.1 Category of density matrices
The canonical definition of category DensityMat (objects — density matrices, morphisms — CPTP channels) and proof of category axioms are in Categorical formalism, §1.
Definition 5.2 (Category of CPTP channels):
This is a well-defined category:
- Composition: is CPTP if and are CPTP
- Identity: — trivial CPTP channel
5.2 φ as endomorphism
Definition 5.3 ( as endofunctor):
induces an endofunctor:
On objects: (identity)
On morphisms: (if is invertible)
Problem: A general CPTP channel is not invertible.
Solution: We consider as an endomorphism in the category with a single object:
Definition 5.4 (Monoid of CPTP channels):
This is a monoid with the composition operation.
— an element of this monoid.
5.3 Relation to monads
Definition 5.5 (Monad of self-modeling):
Consider the functor :
Monad structure:
- Unit (): , (pure state)
- Mult (): , (mixing)
induces a morphism of monads:
Naturality conditions:
Theorem 5.1 (Fixed point as monad algebra):
The fixed point defines a -algebra:
Interpretation: A system in the state of ideal self-knowledge is an "algebra over the self-modeling monad."
5.4 2-categorical structure
Definition 5.6 (2-category of quantum systems QSys):
| Level | Elements |
|---|---|
| 0-morphisms (objects) | Hilbert spaces |
| 1-morphisms | CPTP channels |
| 2-morphisms | Natural transformations between channels |
defines a 2-cell:
(endo-2-morphism of the identity 1-morphism)
Fixed point condition in 2-categorical language:
is an object such that (the 2-morphism reduces to the identity).
6. Corollaries and limitations
6.1 Corollaries of formalization
Corollary 6.1 (Necessity of contraction for ideal self-knowledge):
For the existence of exact it is necessary that be contracting (or have an invariant subspace). Retracted 2026-09-25 [✗]: every continuous on has an exact fixed point (Brouwer); contraction gives uniqueness and geometric convergence of the iterates, not existence. and are not contractions and have exactly one fixed point each.
Corollary 6.2 (Approximate self-knowledge is always possible):
For any continuous and any there exists an -fixed point.
Corollary 6.3 (Relation to thermodynamics):
Contracting CPTP channels correspond to systems with dissipation (attraction to equilibrium).
The fixed point of is the "thermodynamic equilibrium of self-observation."
6.2 Limitations of formalization
Limitation 6.1 (Contraction requirement):
Theorem 3.1 requires . For (isometric ) the fixed point may be non-unique; it always exists (Brouwer; the text said "may not exist" until 2026-09-25).
Limitation 6.2 (Finite-dimensionality):
The proofs use finite-dimensionality of . Generalization to the infinite-dimensional case requires additional conditions (compactness of ).
Limitation 6.3 (Stationarity):
The formalization treats as a fixed operator. In a dynamical system may depend on time: .
Open question: Does a "moving fixed point" exist for ? See Appendix C.
6.3 Physical interpretation
Interpretation 6.1 (Self-modeling as quantum channel):
= CPTP channel means that self-observation:
- Preserves positivity (does not create negative probabilities)
- Preserves normalization (total probability = 1)
- Can decrease information (does not increase distinguishability)
Interpretation 6.2 (Fixed point as self-consistency):
means: "What the system sees coincides with what it is."
This is the state of ideal self-knowledge — the system has no "blind spots."
Interpretation 6.3 (Contraction as humility):
means that each act of self-observation "approaches" the truth.
The system gradually corrects its self-model, converging to an accurate representation.
6.4 Relation to UHM
Relation 6.1 (Reflexive closure):
The condition of self-observation:
is formalized as: for some .
Relation 6.2 (Consciousness):
[Т T-140] includes as a factor.
At : (maximum contribution of integration).
Relation 6.3 (No-zombie theorem):
From interiority hierarchy:
The formalization of ensures: with finite precision.
7. Implementation requirements
This section contains mathematical requirements for implementing the self-modeling operator φ. Concrete architectures and code are the subject of separate specifications.
7.1 Requirements for implementing φ
Requirement 7.1 (Predictive self-modeling operator):
The implementation of must satisfy:
where:
- — contraction parameter ensuring contractivity
- — parameterized map
- — prior state (maximum entropy)
Implementation guarantees:
- Output — valid density matrix (Hermitian, PSD, trace=1)
- Contracting map at
- Differentiability with respect to parameters
Recommended method: Cholesky parameterization guarantees PSD.
7.2 Requirements for sensor encoder
Requirement 7.2 (Encoder: sensors → Γ):
where is a lower-triangular matrix parameterized from sensor input .
7.3 Requirements for action decoder
Requirement 7.3 (Decoder: Γ → actions):
For discrete actions:
For continuous actions:
7.4 Training
Minimization of self-prediction error:
The requirements in this section are sufficient for building a concrete implementation. Cholesky parameterization guarantees correctness of the output density matrices.
8. Operational algorithm for φ
This section provides a concrete algorithm for computing the self-modeling operator φ, suitable for software implementation.
8.1 Algorithm: Basic self-modeling
Input: Coherence matrix
Parameters:
- — contraction coefficient (recommended )
- — anchor weight vector (default )
Algorithm:
FUNCTION φ_basic(Γ, k, w):
# Step 1: Extract diagonal (dephasing in measurement basis)
diag_Γ := diagonal(Γ) # vector of size 7
# Step 2: Build predictive state
P_pred := diag(diag_Γ) # diagonal matrix 7×7
# Step 3: Build anchor state
Γ_anchor := diag(w)
# Step 4: Mix with contraction coefficient
φ_Γ := k * P_pred + (1 - k) * Γ_anchor
RETURN φ_Γ
Guarantees:
- Output — valid density matrix (Hermitian, PSD, trace=1)
- Contracting map with constant
- Computational complexity: where
8.2 Algorithm: Neural network self-modeling
For trainable φ with parameters θ:
FUNCTION φ_neural(Γ, θ):
# Step 1: Vectorize input matrix
x := flatten_upper_triangular(Γ) # 28 parameters (7 diag + 21 coh)
# Step 2: Pass through neural network
h := ReLU(W₁ · x + b₁)
L_vec := W₂ · h + b₂ # 28 parameters for lower-triangular matrix
# Step 3: Reconstruct lower-triangular matrix (Cholesky)
L := unflatten_lower_triangular(L_vec) # 7×7
# Step 4: Build PSD matrix and normalize
Γ_raw := L · L†
φ_Γ := Γ_raw / Tr(Γ_raw)
# Step 5: Apply contraction to anchor
k := sigmoid(θ_k) # trainable coefficient ∈ (0, 1)
φ_Γ := k * φ_Γ + (1 - k) * I/7
RETURN φ_Γ
Training: Minimize next-state prediction error:
8.3 Computing reflection measure R
FUNCTION compute_R_canonical(Γ):
# Canonical definition of R (used in thresholds)
P := Tr(Γ† · Γ) # purity
R := 1 / (7 * P)
RETURN R
FUNCTION compute_Q_phi(Γ, φ):
# Quality of self-modeling (separate measure, see WARNING above)
φ_Γ := φ(Γ)
error := Γ - φ_Γ
error_norm_sq := Tr(error† · error)
Γ_norm_sq := Tr(Γ† · Γ) # = P (purity)
Q := 1 - error_norm_sq / Γ_norm_sq
RETURN Q
8.4 Checking L2 threshold
The function Tr_not_E (partial trace) requires tensor structure. In the minimal 7D formalism () use is_L2_minimal without — see dimension-e.md.
FUNCTION is_L2_conscious(Γ, φ):
# Compute three measures
R := compute_R(Γ, φ)
Φ := compute_integration(Γ) # Σ|γ_ij|² / Σγ_ii²
D_diff := exp(von_neumann_entropy(Tr_not_E(Γ)))
# Check thresholds
RETURN (R ≥ 1/3) AND (Φ ≥ 1) AND (D_diff ≥ 2)
# Minimal version without D_diff (for 7D formalism)
FUNCTION is_L2_minimal(Γ, φ):
R := compute_R(Γ, φ)
Φ := compute_integration(Γ)
RETURN (R ≥ 1/3) AND (Φ ≥ 1)
9. Relation to the regeneration mechanism
The self-modeling operator defines the target state of regeneration: — categorical self-model of the current state [T] (operator φ). For each the self-model is unique (CPTP channel).
9.1 Regeneration as striving toward the self-model
The regenerative term of the evolution equation for is fully derived from the axioms [T]:
where:
- — regeneration coefficient [T] (categorical derivation from adjunction)
- — categorical self-model of the current state [T] (operator φ)
- — unique CPTP relaxation [T] (replacement channel + Bures optimality)
- — V-preservation gate [T] (refines from Landauer, see evolution)
Full derivation: Evolution → Derivation of regeneration form.
Interpretation: The system regenerates by striving toward state — how it "sees itself." Regeneration is an active process of self-realization, where the system becomes its own model.
9.2 Fixed point and viable equilibrium
Theorem 9.1 (Regeneration equilibrium):
At the regenerative term vanishes:
Proof: by definition of fixed point. ∎
Corollary 9.1: At the fixed point the system is in a state of ideal self-knowledge — regeneration is not required, as the current state coincides with the self-model.
9.3 Dynamics outside the fixed point
At a "pull" toward the self-model arises:
Direction of regeneration:
- If : regeneration increases purity
- If : regeneration decreases purity
For regeneration to support viability, it is necessary that:
With an incorrectly constructed the system may regenerate toward a non-viable state. This places constraints on the choice of anchor (see Definition 2.11).
9.4 Relation to reflection measure R
The reflection measure and the regenerative term are related:
Interpretation:
- High (proximity to self-model) → small amplitude of regeneration
- Low (divergence from self-model) → large amplitude of regeneration
A system with good self-knowledge () requires minimal regeneration.
9.5 Stability of viable region
Theorem 9.2 (Regeneration keeps system in ):
Let be a contracting map with fixed point (viable region).
Then at sufficiently large regeneration counteracts dissipation and keeps the system in :
Interpretation: Regeneration is a protective mechanism that uses the self-model as a guide for restoring coherence.
Corrected 2026-09-25 [✗ as stated]. The gate switches regeneration off for , where for every ; the UHM self-models are not contractions (Lemma 2.1, scope). What holds [T], for and at : on the family , is negative below the saddle , positive on and negative above the sink — regeneration protects the window from the saddle up, not from the threshold (living attractor in the window; the admissible in T-336).
9.6 Preservation of positivity under regeneration
The regenerative operator with is a CPTP channel:
with Kraus operators and (from attractor ).
Corollary: Regeneration toward self-model guarantees preservation of:
- Positivity:
- Normalization:
More on CPTP structure of regeneration →
Appendix A: Computation examples
A.1 Depolarizing channel as φ
Fixed point:
Contraction constant:
Reflection measure at fixed point:
A.2 Projection self-observation
Let be an orthonormal basis, .
(Diagonalization in the given basis)
Fixed points:
The set of fixed points is an -dimensional simplex:
where for the Holon.
Remark: This is not a contracting map ( on the set of fixed points).
Appendix B: Proof of CPTP structure preservation
Lemma B.1: If is CPTP and , then .
Proof:
- Hermiticity:
- Positivity:
For any :
(since )
- Normalization:
∎
Appendix C: Generalization to time-dependent φ
Definition C.1 (Dynamic self-modeling operator):
Dynamic fixed point equation:
Theorem C.1 (Existence of dynamic fixed point):
If is contracting with constant for all , and is continuous in , then:
- exists and is unique for each
- is continuous in
- (implicit equation)
Proof: Follows from applying the implicit function theorem in a Banach space.
Octonionic context of self-modeling
In the octonionic interpretation, the self-modeling operator acts on the space . Alternativity of octonions (Artin's theorem [T]) guarantees that is associative when acting on any pair of dimensions, but may exhibit non-associativity when acting simultaneously on three or more dimensions.
This is consistent with the fixed point property : self-consistency is achieved in the full 7-dimensional space where non-associativity is integrated into the structure. Bridge [T] (closed, T15). See structural derivation.
Tensor factorization of φ for composite systems
Tensor factorization of is the key property behind the marginal identity: the regeneration of an autonomous subsystem does not change the unconditioned state of its partner. It does not by itself exclude superluminal signalling: with a Lüders update after a measurement on one side, a state-dependent regeneration on the other side makes its statistics depend on the choice of measurement (Gisin 1990; Polchinski 1991; §8.5 of the page linked). No-signalling of the full dynamics is [C] under the non-selective reading. An earlier sentence here said that factorization guarantees no superluminal channels; retracted.
Preservation of holonomic character. Factorization concerns only the regenerative term . The full dynamics contains:
- (Hamiltonian): creates and preserves entanglement — non-local ✓
- (dissipation): may destroy entanglement, but through common decoherence — non-local in general
- (regeneration via ): local (factorizes) — keeps the partner's unconditioned marginal unchanged
"Holonomy" (the whole > sum of parts) is realized through , not through . Self-modeling () is a local process (each agent models itself, not another). Entanglement is a property of (Hamiltonian dynamics). The theory is not a "local hidden variable theory": only is local, while are non-local.
Refinement: SSB, not gauge freedom. The more precise qualification is spontaneous symmetry breaking (SSB), not gauge freedom:
- Before minimization: -symmetry unbroken, all bases equivalent.
- Upon minimization (T-64 [T] (corrected to the -invariant potential; its vacuum has no sector values — hypothesis (SV))): system "rolls" into a specific vacuum on the manifold of minima . One minimum is selected.
- After SSB: (vacuum stabilizer). Boolean fragment crystallizes as pointer basis fixed by the vacuum.
- Goldstone modes (see goldstone-modes): massless excitations along broken directions .
Analogy: not coordinates in GR, but the Higgs mechanism — generates W/Z masses. In UHM: generates classical objectivity (Dec(Ω) = Boolean logic).
Canonical extension of φ to composite system
Definition (Canonical extension ). For an autonomous holon in a composite system , the extension is defined as:
This is the unique extension compatible with the CPTP structure of and the tensor structure of category .
Theorem: tensor factorization
For a composite system of two autonomous holons and :
Proof:
-
By the definition of autonomy (A1): — conditional independence of and .
-
The operator is defined as left adjoint to the inclusion of subobjects:
- For autonomous subsystems the lattice of subobjects factorizes:
- The left adjoint to the product of inclusions is the product of left adjoints:
Corollary: annihilation of nonlinear contribution
Lemma (Annihilation of regeneration under partial trace). For any CPTP channel and scalar :
Corollary: The regenerative term contributes nothing to — the marginal identity. This is not yet no-signalling of the full dynamics, which fails with the Lüders update and holds only in the non-selective reading [C]; an earlier wording said the term "automatically satisfies the no-signaling prohibition", which is retracted.
Full proof: Physical correspondence — No-signaling prohibition.
Related documents:
- Self-observation — definitions of , and
- Evolution — equation of motion, regenerative term and canonical
- Coherence matrix — definition of
- Viability — purity measure and region
- Unity dimension — integration measure
- Axiom of Septicity — regeneration coefficient
- Interiority hierarchy — levels L0→L1→L2→L3→L4, thresholds
- Categorical formalism — categorical structure of UHM, n-truncations and no-signaling prohibition as natural transformation
- Holon — definition of
- Physical correspondence — No-signaling prohibition — complete proofs