Theorem on Emergent Time
T-53a proves an equivariant bijection of supplied based cyclic label sets; the classifier does not force a clock (§2, §6). PW conditioning assumes a tensor clock and a support constraint (§3.1a, T-87). T-53c proves relative-entropy monotonicity for a specified unital-channel history, with strict loss only under a specified loss/recovery criterion (§7). T-53d is [C under (Fold)+(ClockReadout)] (§8). The finite/depth/history-register constructions of §11 are independent conditional realizations and are retained. A nonlinear regenerator fitted along one trajectory does not give a state-independent linear CPTP law for all inputs.
Contents
- Problem statement
- Time from temporal modality on Ω
- Page–Wootters mechanism for UHM
- Information-geometric time
- Categorical time via ∞-groupoid
- Equivalence theorem
- Arrow of time theorem
- Connection to critical purity
- Corollaries
- Stratificational time
- Precedents and related programmes
- 11.3 Carriers of the aperiodic parameter
- 11.4 The depth register: an exact finite carrier
- 11.5 The de Sitter observer algebra and the holon tower
- 11.6 Direction and the continuous clock
1. Problem statement
1.1 The circularity problem
In the original formulation of UHM, time enters as an evolution parameter:
This is logically circular: dynamics is defined through , but is what we are trying to derive.
1.2 Requirement of Axiom Ω⁷
From Axiom Ω⁷ it follows:
"The ∞-topos is the unique primitive."
Logical consequence: Time must be a function of the structure of category :
1.3 Four levels of the problem
| Level | Problem | Solution |
|---|---|---|
| Kinematic | What is a "moment of time"? | Page–Wootters: correlation with O |
| Geometric | How to measure "the flow of time"? | Bures metric / d_strat |
| Categorical | How to formalize the structure? | ∞-groupoid of paths Exp_∞ |
| Stratificational | What is the arrow of time? | Stratum collapse to T |
2. Supplied clock labels and logical types
The typed kernel uses . Its subobject classifier has : truth values are open subobjects, not rank-one quantum projectors. No seven-atom clock, Hamiltonian, or cyclic action follows from this classifier.
2.1 A chosen cyclic action
Specify a free transitive -set , a generator , and . Then is a bijection [T under the supplied cyclic action]. The hypotheses matter: an arbitrary permutation of order can have smaller orbits and fixed points. is a model choice here.
On the external Boolean algebra , the definition gives an order-preserving Boolean automorphism with . This Boolean algebra is not identified with .
2.2 Cyclic labels and elapsed records
The label is periodic. The history index is directed. The map from the latter to the former is many-to-one, and the label alone cannot record elapsed cycles. Neither index comes with a physical unit until a clock realization and calibration are supplied.
2.3 A compatible finite quantum clock
Choose , an energy basis, and (). Define
The identity proves orthonormality. For ,
Thus the chosen quantum clock realizes the given permutation. Its spectrum and scale are inputs. The logarithm of a unitary is branch-dependent; there is no matrix logarithm of an unspecified logical classifier. No isomorphism between this Hilbert space and “global sections of ” is asserted. The PW tensor factor and support constraint are separate data in §3.1a.
2.4 Logic and the Heisenberg picture
For a fixed linear GKSL generator, the dual on matrix observables is
The dual propagator is unital and completely positive, so it preserves matrix effects [T]. A matrix effect is not a characteristic map . This duality does not unify the topos classifier, selected frame projectors, clock rates, and dissipation. A nonlinear state flow has no single input-independent linear Heisenberg dual.
2.5 Predicates on the cyclic labels
For , one may define and . These are logical operations relative to the supplied finite action [D]. They supply no quantum evolution or irreversible arrow.
2.6 Separate constructions
The label set yields a permutation and, after Hilbert-space realization, a unitary clock step. The chosen state-space site yields a logical classifier. A chosen numerical generator yields a state trajectory. Connecting these constructions requires typed maps and additional compatibility data; none of these arrows is forced by merely naming .
2.7 Scope of the temporal language
Theorem 2.7.1. A chosen generator on a free transitive cyclic set gives the action in §2.1. A general -topos specifies neither such a generator nor a clock–system tensor factorization.
3. Page–Wootters mechanism for UHM
Both the clock tensor factor and are explicit model inputs. A direct-sum algebra, the logical classifier, or stationarity alone does not supply them. The conditional scope and the mixed-state counterexample are given in §3.1a.
3.1 The idea of the mechanism (standard formulation)
In quantum gravity, the following construction is used:
Full system:
- — clock subsystem
- — the rest of the system
Wheeler–DeWitt condition:
Time arises as correlation between the clock and the system.
3.1a Page–Wootters: supplied tensor structure and constraint
T-87 is [C under a specified clock tensor factor and support constraint]. Specify a Hilbert space , another space , their tensor product, self-adjoint , and with
These are declared model data. A Wedderburn direct-sum decomposition, a clock-labelled O-axis, or the classifier does not force a tensor factor. In particular the quotient of a Hilbert space by a subspace is not a complementary tensor factor.
The support condition implies . The converse fails: for distinct constraint eigenvalues commutes with , yet for every scalar shift . Pure stationarity selects one energy eigenspace, which can be shifted to zero; mixed stationarity does not.
For supplied clock projectors , let and define
Positivity and trace normalization follow directly. A compatible noninteracting constraint can yield unitary conditional dynamics, as in the original PW construction; no universal conditional law follows merely from stationarity. Interaction corrections require explicit estimates, not a bare . The finite history construction in §11 is separate and is retained.
3.2 Adaptation for UHM
In the 7D structure of UHM, the natural candidate for the role of a clock is dimension O (Foundation).
Justification:
- O — connection to the quantum vacuum
- O participates in regeneration: (see categorical derivation of κ₀)
- Physically: O is the "source" feeding the dynamics
3.3 Formal construction
Type convention. The tensor product below is a separately supplied joint system: if and , its dimension is 42. The O-labelled one-dimensional subspace of the original is not this clock factor; no factorization of that seven-dimensional space is asserted. Conditional states require .
Step 1: Decomposition of Γ
where .
Step 2: Page–Wootters constraint
where the constraint operator:
Step 3: Conditional state
Internal time is defined via conditional states:
where:
- — basis of eigenstates of clock O
- — normalization
3.4 Page–Wootters theorem
Let satisfy the constraint . Then the conditional states evolve according to:
where is the effective Hamiltonian arising from .
Corollary: Time is not an external parameter, but a parametrization of correlations within the global state .
The "corrections" are not small in general. For the conditional states are related by a unitary step, . When the constraint couples clock and system, A. R. H. Smith and M. Ahmadi proved for clocks with continuous spectrum that the conditional state obeys a time-nonlocal Schrödinger equation, in which the system Hamiltonian is replaced by an integral operator ("Quantizing time: interacting clocks and systems", Quantum 3, 160 (2019), arXiv:1712.00081); the local generator of §3.6 is at best the leading term in , and the corpus has not shown that the seven-level clock escapes the nonlocal form.
The conditional evolution of an interacting clock depends on its joint Hamiltonian and reading instrument. A cohesive or flat modality alone supplies no estimate on the interaction correction. Clock model
3.5 Clock basis for 7D
For :
where are eigenstates of .
3.6 Explicit constructions for UHM
Complete formulas for the 7D UHM system are defined in the respective master documents:
| Construction | Formula | Master definition |
|---|---|---|
| Clock Hamiltonian | dimension-o#гамильтониан-часов-h_o | |
| Shift operator | dimension-o#оператор-сдвига-v_o | |
| C*-algebra of clocks | dimension-o#c-алгебра-часов-a_o | |
| Interaction Hamiltonian | axiom-omega#гамильтониан-взаимодействия | |
| Full constraint | axiom-omega#свойство-2 | |
| Effective Hamiltonian | evolution#вывод-h_eff |
The last row is exact only for ; with interaction it is the leading term of a time-nonlocal law (§3.4).
3.7 Discreteness of time for finite systems
For time is fundamentally discrete, not continuous.
Question: If τ ∈ ℤ₇ is discrete, why does the evolution equation use dΓ/dτ (a derivative)?
Answer:
- Minimal formalism (N=7): τ is discrete, equations are difference equations (Δτ instead of dτ)
- Macroscopic limit (N → ∞): τ approaches a continuum, equations are differential
- Practice: The differential form is a convenient approximation when Δτ ≪ the characteristic timescales of the system
For implementations: Use the discrete form: Γ(τ+1) = Γ(τ) + Δτ·(...) with step Δτ = 2π/(7ω₀).
is the dimensionality of the clock Hilbert space, not the cardinality of the set of moments. The distinction:
- Clock basis: 7 orthogonal states — basis of , analogous to 7 divisions on a clock face
- Moments of time: — a cyclic group. The system passes through cycles indefinitely, like clock hands with 7 divisions
- Chronon: — the minimal quantum of subjective time, determined by the characteristic frequency of the system, not by the number 7
For identical seven-level clocks the joint Hilbert space has dimension . The summed generator has distinct readings and period . Composite clocks
Theorem (Discreteness of time): For a finite-dimensional system with , the internal time takes values from the cyclic group:
For UHM with :
Corollaries:
| Property | Discrete time () | Continuous limit () |
|---|---|---|
| Set of times | (7 moments) | or |
| Topology | Discrete, cyclic | Continual |
| Chronon (minimal quantum) | ||
| Fundamental group | ||
| Evolution equation | Difference | Differential |
Interpretation:
- Quantization of the present: There exists a minimal "quantum" of subjective time — chronon
- Cyclic time: Time locally has the structure of , not
- Emergent continuity: Continual time is the macroscopic approximation for
3.8 Limit N → ∞ and connection to physics
As , the discrete time transitions to continuous time algebraically, not topologically.
Topological error: topologically!
- Projective limit — totally disconnected space
- — connected space
- They are topologically distinct
Correct formulation of the limit:
Definition (Scaled limit):
This is a scaled limit, not a topological one.
Theorem on algebraic limit
As , the group algebra converges to the algebra of continuous functions on the circle:
as C*-algebras (algebraically, not topologically).
Proof:
(a) Structure of the group algebra:
(b) Fourier transform:
Isomorphism :
(c) Limiting transition:
As , the spectrum becomes dense in :
(d) C-isomorphism:*
By the Gelfand–Naimark theorem:
∎
Chronon as a function of N:
| N | Interpretation | |
|---|---|---|
| 7 | UHM chronon (minimal quantum of subjective time) | |
| 100 | Mesoscopic limit | |
| 0 | Classical limit (continuous time) |
Correspondence theorem (classical limit)
For any observable :
where .
Proof:
Average over discrete time:
As with (where ):
∎
Corollary for UHM:
Classical continuous time on a circle is the macroscopic approximation of discrete internal time: the readings become dense in , whose circumference does not change. A line is not obtained this way (see the end of this subsection).
Theorem (Continuous limit — algebraic):
In the limit at fixed :
- (chronon vanishes)
- fills the circle of circumference ; the period does not grow
- Algebraic convergence: (group algebras, not groups!)
Key clarification: The transition is algebraic (group algebras ), not topological ().
Theorem on composite clocks and continuous limit
Summed clocks. The number of distinct readings is set by the generator. Its degeneracies separate the dimension of the joint register from the number of clock readouts.
What holds instead (elementary; checked numerically for ). For holons with identical clocks the summed clock has distinguishable readings, period and finest orthogonal resolution
which shrinks like , not like . As the readings become dense in a circle of fixed circumference — the algebraic limit above — but the circle does not unroll into a line: composite O-clocks do not supply an aperiodic time (§11.2).
Positional clock. If the registers are digits , a successor with carry traverses ordered readings. This requires the positional update and the corresponding history constraint. Depth register
Let be the logical Liouvillian with . Then the discrete evolution converges to the continuous Lindblad equation:
For holons with the resolution of the summed clock the per-step error is of order : small polynomially, not exponentially.
Proof: Standard estimate via Taylor formula for the exponential: .
Substituting :
As this tends to zero like .
Physical interpretation:
| System | M | Readings | Continuity | |
|---|---|---|---|---|
| Single holon | 1 | 7 | Discrete | |
| Neuron ( molecules) | Quasi-continuous, on a circle | |||
| Macroscopic system | Continuous circle of circumference , not |
Connection to the chronon:
| Scale | Chronon | Time |
|---|---|---|
| Subjective (N = 7) | Discrete, | |
| Neural (N ~ 10⁸) | Quasi-continuous | |
| Physical (N → ∞) | Continuous circle for the summed O-clock; from the depth register (§11.4) |
Corollary for interpretation:
At fixed the limit of the summed-clock readings is a circle of circumference . The line arises under the separately defined depth-register scaling about an interior reading. T-118
- Discreteness of states of consciousness
- Finite information capacity
- Topology of ∞-groupoid
Discreteness of time leads to a discrete ∞-groupoid instead of a continuous one. See Categorical formalism.
4. Information-geometric time
4.1 Bures metric
The space of density matrices has a natural Riemannian structure.
where is the solution of the Lyapunov equation:
Explicit formula for the distance (Bures angle):
where — fidelity.
4.2 Geometric time
Between two configurations and , the information time:
where the infimum is taken over all paths connecting and .
4.3 Flow of time
Let be a continuous family of states. The speed of flow of internal time:
Interpretation: "The flow of time" is the rate of change of Γ in the Bures metric. Time "flows faster" when Γ changes more.
4.4 Correspondence with dynamics
For unitary evolution with :
For close to a pure state :
Corollary: The time-energy uncertainty relation:
is derived from the geometry of the state space, not postulated.
5. Categorical time via ∞-groupoid
5.1 ∞-groupoid of experiential paths
∞-category is defined as:
0-cells (objects):
(History Hist is not included — it is derived as the structure of the ∞-groupoid)
1-morphisms:
2-morphisms:
n-morphisms:
5.2 Time as a 1-morphism
Time is a 1-morphism in :
Direction of time — choice of orientation on 1-morphisms.
Equivalent moments of time — 2-isomorphic 1-morphisms.
5.3 Theorem on internal time
In the ∞-groupoid :
- History — automatically arises as the loop space:
- Temporal structure — homotopy type:
- Arrow of time — orientation σ on 1-morphisms.
5.4 ∞-topos of sheaves
∞-topos — category of ∞-sheaves on :
- ∞-topology: Cover = family of paths covering a neighborhood
- ∞-sheaf: Functor , satisfying the descent condition
is an ∞-topos and has:
- Internal logic: Homotopy type theory (HoTT)
- Internal time: Modality of type "in the future", "in the past"
- Subobject classifier: ∞-groupoid of truth values
Corollary: The logic of experiential content is temporal modal logic, derivable from the internal structure of the ∞-topos.
6. Label correspondences and their limits
6.1 Three different types
The PW label is a clock reading. Bures length is a functional of a state curve. A categorical history is a functor from a specified index category. These are not automatically the same object, nor do they supply physical time units.
6.2 T-53a: supplied based cyclic torsors
Theorem. Let and be given free transitive -sets with chosen generators and basepoints. There is a unique equivariant bijection satisfying , namely
Proof. Freeness makes the label unique modulo , so the formula is well-defined. Transitivity gives every point this form. The inverse exchanges the indices 1 and 2; equivariance is immediate. Without a chosen image of , there are exactly choices, hence no preferred bijection.
This is T-53a [T for the supplied label sets]. Realizing three proposed descriptions as such torsors is additional data; the theorem does not prove equivalence of their processes, metrics, or dynamics. No torsor is forced by the classifier .
Step 1: Bures length of a supplied curve
For a curve of full-rank states, the unique symmetric logarithmic derivative solves . For the Bures convention , the infinitesimal metric is
Lemma 6.1. On an interval with , arc length is a strictly increasing local parameter, proportional to time exactly when speed is constant on the interval. A stationary curve has zero arc length.
Step 2: Additional categorical realization
Lemma 6.2 (scope). A state-curve assignment to paths requires an observation map. Geodesicity, faithfulness and preservation of distance need separate verification. A path groupoid has invertible arrows, while a directed history also admits irreversible processes.
Step 3: Directed depth is not a cyclic torsor
Lemma 6.3. The map loses the cycle count. If then has inverse . An irreversible channel cannot implement this cyclic action as an endomorphism. Loss and recovery
Step 4: Exact conclusion
The equivariant label bijection is a theorem about supplied cyclic label sets. Neither their physical calibration nor a universal equivalence of PW, Bures, categorical, and stratal time is derived. T-53d is independently conditional on a supplied fold and clock readout in §8.
7. Arrow of time theorem
T-53c concerns the monotonicity of a specified history under unital CPTP steps. An identity channel gives equality throughout.
7.1 Categorical formulation
For any path γ: [0,1] → in state space:
where:
- , if the path is induced by a unital CPTP channel,
- , if the path requires inverting such a channel
For an arbitrary CPTP channel the monotone quantity is the relative entropy to a fixed point: if , then .
Proof:
Relative entropy does not increase under any CPTP map: (Lindblad 1975; Uhlmann 1977). If is unital, then , and with the inequality reads , that is,
The dissipator of UHM has Hermitian Lindblad operators (the pointer projectors) and is therefore unital, so the linear part of the evolution raises (§10.4, part 1).
A non-unital channel can lower entropy. For example, sends to a pure state. Unitality supplies the fixed comparison state in the data-processing proof. A nonlinear flow needs its own monotonicity proof.
∎
7.2 Physical interpretation
Unital steps in fixed dimension cannot decrease entropy. External preparation, cooling and other non-unital operations require their own energy and entropy accounting.
7.3 Connection to regeneration
Regeneration locally decreases entropy, but only when:
Total entropy (system + energy source) grows:
Corollary: The gate in the regenerative term (refining from Landauer) is not a postulate, but a consequence of the CPTP structure, thermodynamics and V-preservation.
7.4 ∞-categorical resolution
Typed replacement for T-53c [D]+[T under unitality]. A directed history is a functor : its objects are and its arrows are specified channels and their composites. This is a category object, not a claim that every topos arrow points to its terminal object. If the steps are unital, is non-increasing by data processing (§7.1).
Record numbering alone proves no physical irreversibility: identity and unitary channels also admit directed records. A common inverse is excluded when the selected action has no CPTP inverse; strict loss on specified states must be established separately. A verified example is depolarization , , at in T-53c. Nonequivalence of a functor has no universal linear-kernel criterion.
If decreases strictly, no CPTP map recovers both and from their outputs: data processing would give the reverse inequality. Preparing one known is a different operation.
The condition gives the terminal mapping space. An equilibrium or universal time parameter would require further dynamical input.
A selected Hessian score and the limits of agency claims
Choose a potential on a -dimensional manifold and a critical point . Define [D]
Rank-nullity gives [T]. At a critical point, the Hessian transforms by congruence under an invertible coordinate change, so its nullity is well-defined. On the full trace-one Hermitian tangent space of , , hence the range is to ; a separately chosen six-dimensional diagonal model has bound .
Under the Morse–Bott hypothesis, the Hessian kernel is the tangent space of the critical manifold, so the score is one plus its local dimension. Without this hypothesis, zero quadratic modes need not produce alternatives: has zero Hessian at its sole local minimum. No potential, stationary point, or universal score at is supplied by terminality.
The terminal mapping property fixes no numerical bound on a set of action representatives. Such a count requires a specified set and equivalence relation.
8. Purity bounds and a conditional fold law
8.1 What purity actually bounds
For a density matrix , let and . Then and
This proves a kinematic upper bound [T]. It is not a Bures-speed formula and yields no positive lower bound or viability equivalence. Any energy-diagonal state commutes with regardless of its purity. For , the diagonal state with has and ; the same zero speed holds on both sides of that purity. Thus no distinguished temporal transition follows at .
For the separately chosen rank-one Hamiltonian in a fixed frame, direct matrix multiplication gives
This is a Hilbert–Schmidt identity for that Hamiltonian, not the Bures norm, not a general multilevel clock formula, and not a clock factorization theorem.
8.2 T-53d under (Fold)+(ClockReadout)
Use the explicit hypotheses (Fold) and (ClockReadout). Assume a finite-dimensional field at an interior equilibrium has one simple zero eigenvalue, all other eigenvalues with negative real parts, and nonzero fold coefficients with . In a smooth center coordinate and the specified time parameter,
Set and . The equation extends to with value . Its two roots have derivative . The implicit function theorem gives , hence
Choose the branch with . Differentiating the smooth reduced field gives
The transverse eigenvalues stay strictly in the left half-plane, so this is the slow relaxation eigenvalue. The relaxation time diverges as . This statement concerns relaxation around an equilibrium, not the speed of that equilibrium, which is zero.
Independently select a clock-rate readout that is nonnegative along the chosen branch and satisfies . Substitution gives
This proves T-53d [C under (Fold)+(ClockReadout)]. The two hypotheses must be verified for the selected UHM numerical model; they do not follow from primitivity of its fixed linear GKSL part, purity, or categorical terminality. A nonzero readout at the fold need not slow down; a readout with vanishing first derivative can have a different exponent. The normal-form framework is in Kuznetsov’s notes, and sensitivity of scaling to degeneracy is studied by Kuehn (2009).
To express this as further assume a positive linear calibration , . If instead is linear in , the clock rate is linear in . No universal or infinite subjective duration follows.
The purity bound applies kinematically. The square-root readout law requires both the fold hypothesis and a calibrated clock readout.
9. Corollaries
9.1 Modification of the evolution equation
A finite cyclic clock repeats its outcome after one period. If conditional states depended on this outcome alone, a trajectory with strictly decreasing distinguishability could not continue around the cycle. A depth register records the additional step index.
Form in an aperiodic parameter [T] (relative to the depth register, §11.4):
where:
- — the aperiodic semigroup parameter; at finite resolution is the reading of the depth register (§11.4), and the Page–Wootters tick is its lowest digit (§10.3)
- — effective Hamiltonian from constraint , exact for and the leading term otherwise (§3.4)
- The dissipator and the regenerator are the axiomatic dynamics in (evolution); what is derived is that a finite timeless world reproduces them as conditional states relative to the depth register — the linear part as a state-independent law (Theorem 11.1), the full flow along each solution (Theorem 11.3)
9.2 Extended role of dimension O
Dimension O now has a dual role:
- Energy source: Provides for regeneration
- Internal clock: Parametrizes internal time via the Page–Wootters mechanism
9.3 Extended categorical structure
G F
DensityMat_C ──────────► DensityMat ────────────► Exp
│ │ │
│ constraint │ CPTP │ induced
▼ ▼ ▼
DensityMat_C ──────────► DensityMat ────────────► Exp
↓ embed
Exp_∞ (∞-groupoid)
↓ sheafify
Sh_∞(Exp) (∞-topos)
where:
- DensityMat_C — category with Page–Wootters constraint
- G — functor "conditional states"
- Exp_∞ — ∞-groupoid of paths
- Sh_∞(Exp) — ∞-topos of sheaves
9.4 Experimental predictions
| Prediction | Formula | Theor. status | Exp. status |
|---|---|---|---|
| Time slowdown at decoherence | [T] Corollary of T.8.1 | Requires verification | |
| Discreteness of internal time | [T] Corollary of §3.7 | Requires verification | |
| Temporal entanglement | even when | [T] Corollary of P-W | Requires verification |
- Theor. status [T]: Prediction is mathematically derived from the UHM formalism
- Exp. status: Prediction requires experimental verification
10. Stratificational time
10.1 Base space as nerve of category
From Axiom Ω⁷ the base space is defined as:
where is the nerve of the category of Holons.
10.2 Stratification of X
Space X is stratified:
where:
- — terminal object (attractor Γ*)
- — edges (morphisms to T)
- — n-simplices
10.3 Temporal stratification: two indices, one arrow
The cyclic tick labels a selected clock; the directed index numbers records in a specified history. Relating them by the convention is neither a bijection nor a physical derivation of elapsed time from a periodic clock. A non-increasing functional on a cycle is constant.
Record numbering is not automatically a stratification of and proves no inequality . Geometric stratification needs separate data; terminality does not provide it.
10.4 Arrow of time theorem (stratificational)
For specified unital CPTP steps with ,
For a fixed unital GKSL generator , apply this to , : . Primitivity is an additional condition for convergence to , not a consequence of terminality. Strictness requires a separate check of the action on the selected state; always gives equality.
Proof. Data processing gives and unitality gives . The identity converts it into non-decreasing entropy. This says nothing about stratum dimension or convergence of every unital history.
For a nonlinear flow , monotonicity and attraction require separate proofs. Realizing a specified history in the depth register does not by itself select a cosmic clock.
10.5 Connection to thermodynamics
| Mathematical object | Scope of thermodynamic inference |
|---|---|
| Specified unital history | is non-increasing [T] |
| Strictly information-losing step on a declared class | Irreversibility under a specified pair/class recovery criterion with no common recovery proved to exist [C] |
| Full nonlinear flow | Needs a separate functional and environment model; T-53c supplies neither |
| Terminal topos object | Determines no physical attractor or arrow |
10.6 Stratified metric
Definition (Metric d_strat):
where:
- γ — path through strata
- ds_α — Connes metric on stratum S_α
Theorem 10.2: d_strat is consistent with the Bures metric:
The following constructions take a clock factor, a reading instrument and a global support constraint as inputs.
11. Precedents and related programmes
The Page–Wootters mechanism on which §3 rests is more than forty years old, and it has a literature that this page did not cite: the classic objections to it, answers to those objections, experiments, and results about the kind of clock UHM uses — a finite, periodic one. This section says what each work established, how it stands and on whose judgment, which UHM construction it bears on, and where UHM differs; §11.2 then checks the clock and the arrow of §10 against these results. Every mapping between UHM and these works is an interpretation [I] unless a theorem is named.
Terms used below. A constraint (an equation of Wheeler–DeWitt type) says that the global state is annihilated by the total Hamiltonian, so nothing in it changes with an external time. The conditional state is the state of the rest of the system given a clock reading (Definition 3.1). An ideal clock has a Hamiltonian whose spectrum is the whole real line and a reading that runs monotonically; a periodic clock returns to its initial state after a fixed period. A relational observable is a gauge-invariant quantity of the form "the value of when the clock reads ".
11.1 The mechanism, its objections and the answers
- Page and Wootters (1983); Wootters (1984). D. N. Page, W. K. Wootters, "Evolution without evolution: dynamics described by stationary observables", Phys. Rev. D 27, 2885–2892 (1983); W. K. Wootters, "'Time' replaced by quantum correlations", Int. J. Theor. Phys. 23, 701–711 (1984). The universe is in a stationary state; one subsystem serves as a clock, and the state of the rest conditioned on a clock reading evolves by the Schrödinger equation. Wootters argued that coordinate time is unobservable while clock time is observable, so every statement about evolution can be replaced by a statement about correlations between a clock and another system, all stored in one timeless state. Standing: the starting point of the relational-time literature below; C. Marletto and V. Vedral (2017) call the model elegant but write that it "has never been developed further, because it was criticised for generating severe ambiguities". For UHM: §3.3 adopts the mechanism as it stands; UHM's additions are the choice of clock — the seven-dimensional O-register — and the claim that the split is derived (T-87). One step of that claim needs an extra assumption. Page and Wootters take the global state to be a stationary state, that is, an eigenstate of the total Hamiltonian, and the constraint form follows after the energy is shifted to zero. For a mixed global state, stationarity gives only , whereas the constraint of §3.3 requires the state to lie in the zero eigenspace of ; §3.1a used to state that the constraint "follows from stationarity" without this assumption; it now names the assumption, and step 4 of T-87 is [C].
- Kuchař (1992). K. V. Kuchař, "Time and interpretations of quantum gravity", in Proceedings of the 4th Canadian Conference on General Relativity and Relativistic Astrophysics, eds. G. Kunstatter, D. Vincent, J. Williams, World Scientific 1992; reprinted in Int. J. Mod. Phys. D 20, Suppl. 1, 3–86 (2011). The classic review of the problem of time. Its objections to the conditional-probability reading of Page–Wootters, as paraphrased by Höhn, Smith and Lock (2021, §VIII C): (1) for a relativistic particle it gives a wrong localisation probability; (2) conditioning on a clock reading uses operators that do not commute with the constraint, so it appears to violate the constraint; (3) it gives wrong two-time propagators — after one conditioning the clock is "stuck" and time does not flow.
- Unruh and Wald (1989). W. G. Unruh, R. M. Wald, "Time and the interpretation of canonical quantum gravity", Phys. Rev. D 40, 2598–2614 (1989). They proved that in ordinary Schrödinger quantum mechanics, for a system whose Hamiltonian is bounded below, no dynamical variable can correlate monotonically with the Schrödinger time parameter: a clock made of such a system is never ideal. Standing: an accepted theorem; Höhn, Smith and Lock (2021, §III) present it as the refinement of Pauli's observation that no self-adjoint time operator is canonically conjugate to a bounded Hamiltonian. Its consequence for UHM is taken up in §11.2.
- Relational evolution. Gambini, Porto and Pullin computed clock-conditioned probabilities for a specified clock model. A clock-dependent decoherence estimate requires its measurement resolution and dynamics. Gambini, Porto and Pullin (2004)
- Experiments. E. Moreva, G. Brida, M. Gramegna, V. Giovannetti, L. Maccone, M. Genovese, "Time from quantum entanglement: an experimental illustration", Phys. Rev. A 89, 052122 (2014), arXiv:1310.4691: two polarization-entangled photons; the polarization of one serves as a clock with two readings, the other photon evolves in birefringent plates; an "observer" who reads the clock photon sees the other evolve, a "super-observer" who measures only global properties finds the joint state static. E. Moreva, M. Gramegna, G. Brida, L. Maccone, M. Genovese, "Quantum time: experimental multitime correlations", Phys. Rev. D 96, 102005 (2017), arXiv:1710.00707: two-time correlations in the Page–Wootters picture and a violation of a Leggett–Garg inequality for the internal observer, with the clock encoded in the position of a photon. Standing: the authors call these illustrations — they show the conditional-probability structure in a prepared state, not the absence of an external time. For UHM: the clocks used had two readings (2014) or a position register (2017); the seven-level O-register has not been realised in any experiment.
- Clock ambiguity. A. Albrecht, A. Iglesias, "The clock ambiguity and the emergence of physical laws", Phys. Rev. D 77, 063506 (2008), argued that different choices of the clock subsystem make one timeless state describe different laws. C. Marletto, V. Vedral, "Evolution without evolution and without ambiguities", Phys. Rev. D 95, 043510 (2017), arXiv:1610.04773, answered that a good clock is a subsystem with many distinguishable states that ideally does not interact with the rest, , and that if such a split yields unitary evolution with a time-independent Hamiltonian it is unique. Standing: a published answer, valid under its stated assumption of a non-interacting clock. For UHM: the constraint couples the O-register to the rest through (Property 2) and the clock has seven states, so this uniqueness argument does not cover it; the corpus rests the uniqueness of the clock register on its own argument (T-87, step 3).
- Interacting clocks. With a clock–system interaction, the exact conditional equation can be nonlocal in time. A local effective generator requires additional approximation hypotheses. Smith and Ahmadi (2019)
- Quantum reference frames. F. Giacomini, E. Castro-Ruiz, Č. Brukner, "Quantum mechanics and the covariance of physical laws in quantum reference frames", Nat. Commun. 10, 494 (2019), arXiv:1712.07207: reference frames attached to quantum systems transform into one another by a "superposition of coordinate transformations", and entanglement and superposition are frame-dependent. Standing: an active programme; Höhn, Smith and Lock extend it to changes of temporal frame (2021, §VII). For UHM: the O-register is a temporal reference frame, and the corpus fixes one such frame without defining a change of clock; statements about entanglement between the clock and the rest, including the "temporal entanglement" of §9.4, hold relative to that frame.
11.2 The finite periodic clock, Unruh–Wald, and the arrow
- The Salecker–Wigner–Peres clock: prior art for the O-clock. H. Salecker, E. P. Wigner, "Quantum limitations of the measurement of space-time distances", Phys. Rev. 109, 571–577 (1958); A. Peres, "Measurement of time by quantum clocks", Am. J. Phys. 48, 552–557 (1980). The standard finite quantum clock: equally spaced levels , period , time states obtained from the energy states by the discrete Fourier transform, which rotate into one another every and return after steps; between ticks a time state spreads over its neighbours, and no finite clock satisfies the canonical commutation relation with its Hamiltonian. M. P. Woods, R. Silva and J. Oppenheim call it "the model of choice for finite-dimensional clocks" ("Autonomous quantum machines and finite-sized clocks", Ann. Henri Poincaré 20, 125–218 (2019), arXiv:1607.04591, Appendix B). The O-clock of §3.5–3.6 — , Fourier clock states , cyclic shift , chronon — is exactly this clock with . The discreteness of its readings, the cyclic group and the resolution are properties of every such clock, so the "discreteness of internal time" listed in §9.4 is a property of the chosen clock rather than an independent test of UHM; what is specific to UHM is the claim that the clock has exactly seven levels and is the O-dimension.
- Unruh–Wald against the O-clock: no collision at the level of the clock. has the finite spectrum , bounded below and above, so the Unruh–Wald theorem applies: no observable of the O-register correlates monotonically with time. The corpus does not claim otherwise — §10.3 states that no function of the tick alone can be monotone. Giovannetti, Lloyd and Maccone (2015, §C) spell out the consequence for Page–Wootters: a non-periodic time needs an unbounded Hamiltonian, "a system with finite global energy will have periodic evolution", and in that case "except as an approximation internal observers will not be able to use a Schrödinger equation".
- Periodic clocks: where the collision lies. L. Chataignier, P. A. Höhn, M. P. E. Lock, F. M. Mele, "Relational dynamics with periodic clocks", New J. Phys. 28, 034504 (2026), arXiv:2409.06479, extended the trinity to periodic clocks and proved that relational observables relative to a periodic clock are invariant only if the quantity itself is periodic — "counting winding numbers does not lead to invariant observables relative to the periodic clock" — and that the dynamics relative to a periodic clock is "necessarily periodic" in all three formulations, Page–Wootters included; a system that evolves periodically relative to a periodic clock can still evolve monotonically relative to an aperiodic one. Höhn, Smith and Lock (2021, §III A) had already noted that with a periodic clock one must keep track of its winding number to follow a non-periodic evolution. This bears on UHM in two places: A cyclic clock registers a phase within a period. Strict irreversible relaxation requires an additional reading that distinguishes successive periods, for example the depth register. The argument is purely operational: identical clock outcomes must receive the same conditional state if those outcomes are the only conditioning data.
11.3 Carriers of the aperiodic parameter: options
Proven [T] (proofs below, witnesses in website/scripts/check_core_numbers.py): in a world with pure point spectrum no clock whose readings are the orbit of a unitary group — continuous or discrete, periodic or not — supports dissipation on its whole orbit, and every Page–Wootters conditional dynamics is linear in the global state up to normalisation. Verdict: the O-clock, the stratal depth without a register, and composite O-clocks with incommensurate frequencies do not carry the parameter of T-53b; an ideal clock register with an infinite environment carries it approximately (Davies limit); the depth register of §11.4 — a finite chain of readings, not an orbit — carries it exactly, and T-53b holds there as a theorem [T]. An intermediate version of this box, of the same day, stated as a verdict that only the ideal clock with an infinite environment makes T-53b true, as [C], and left the choice of carrier to the author [Pr]; that verdict is superseded by §11.4.
T-53b asks three things of its time parameter: (R1) the dynamics in it is not periodic; (R2) a Lyapunov functional decreases strictly along it (Theorem 10.1); (R3) the nonlinear regenerative term acts in it.
[T] Lemma (almost periodicity). Let the clock readings be the orbit of a unitary group — , , or , — and let the joint generator of the clock and everything it is correlated with have pure point spectrum; in particular let the world be finite-dimensional, the clock periodic or not. Then every conditional expectation value is an almost periodic function of the reading, and a non-increasing almost periodic function is constant. Hence (R2) fails for every such clock on its whole orbit.
Proof. An expectation value is (or for the eigenphases of ), a trigonometric polynomial, or a uniform limit of such sums in the pure-point case; both are almost periodic in the sense of Bohr, on or on . For an almost periodic and every the -almost periods are relatively dense. If for some , take and an -almost period : then and , so is not non-increasing. This is the mechanism of the quantum recurrence theorem (P. Bocchieri, A. Loinger, Phys. Rev. 107, 337 (1957)). Witness: a closed seven-level system with an eight-level environment and a random Hamiltonian, started in a pure product state: the relative entropy falls from to for and afterwards rises again by up to .
[T] Lemma (conditional dynamics is linear). In the Page–Wootters construction the conditional state at the clock reading is — a linear map of the global state followed by normalisation, and is constant for an ideal clock that does not interact with the rest. For interacting clocks the equation becomes time-nonlocal (Smith–Ahmadi 2019, §11.1) and stays linear in the global state. Hence is not the conditional dynamics of any single constraint for all initial states: the map from the initial state to the conditional history is linear, the regenerative flow is not. It can enter as a law postulated in ; each of its solutions separately is a conditional history of a constraint fitted to that solution (§11.4, Theorem 11.3); or it can enter as an effective equation — for example the mean-field limit of many exchangeable systems (H. Spohn, "Kinetic equations from Hamiltonian dynamics: Markovian limits", Rev. Mod. Phys. 52, 569 (1980)). Where varies, the conditional dynamics is normalised-linear, the Rembieliński–Caban class, which by physics correspondence, §8.7 cannot carry the viability gate either.
Proof. Conditioning, partial trace and the dynamics of the global state are linear; only the division by is not.
The candidates.
| Carrier | (R1) aperiodic | (R2) dissipation | (R3) | Verdict for T-53b |
|---|---|---|---|---|
| O-clock | no: period seven ticks (Chataignier–Höhn–Lock–Mele 2026) | no | no | [✗] |
| Stratal depth | not an observable relative to the O-clock: a winding number, which is not an invariant relational observable relative to a periodic clock (§11.2) | only through a register that records it | only through that register | not a carrier by itself; recorded in a register it is the carrier of the last row |
| Composite O-clocks with incommensurate frequencies | yes: the orbit on the torus never closes | no, by the almost-periodicity Lemma: two clocks with never return exactly, yet at the clock state overlaps its initial state to | no | [✗]; the readings form — by Kronecker's theorem the orbit is dense in the torus — not , so T-118 is not obtained this way either |
| Ideal clock register ( on , spectrum ; or a counter on ) | yes | only together with an infinite environment with continuous spectrum, in a Markov limit that gives approximately (E. B. Davies, "Markovian master equations", Commun. Math. Phys. 39, 91 (1974)) | postulated (Lemma above) | approximate: only in the weak-coupling limit, with an infinite environment; superseded by the last row |
| Semigroup parameter taken as primitive | yes | as postulated | as postulated | the same as the ideal clock: by the trinity of Höhn, Smith and Lock (2021) the conditional dynamics relative to an ideal clock is the Schrödinger picture in |
| Depth register: readings as a chain, realised positionally in O-registers, with a two-holon environment and a Feynman–Kitaev constraint (§11.4) | yes: no reading follows , the chain is not a group orbit | yes, exactly at every reading, for the linear part as a state-independent law (Theorem 11.1) | along each solution (Theorem 11.3) | [T]: T-53b holds, in a world of dimension |
Cost of the ideal clock. (i) An ideal clock has a Hamiltonian unbounded below, so no ground state; by Unruh and Wald (1989) no clock with a Hamiltonian bounded below reads time monotonically. (ii) Its finite approximations — the quasi-ideal clocks of M. P. Woods, R. Silva and J. Oppenheim, whose back-reaction is exponentially small in the clock dimension at a linear cost in energy (Ann. Henri Poincaré 20, 125 (2019)) — are finite, hence periodic, and by the almost-periodicity Lemma support dissipation only over a finite window; exact T-53b needs . (iii) The environment must be infinite, and is then a weak-coupling approximation, not an exact law. (iv) The time of the dynamics is not emergent from the O-dimension: the O-clock becomes a kinematic label read off (, ), and the arrow of Theorem 10.1 is monotone in , as it is already stated. (v) is postulated in , and the measurement reading of physics correspondence, §8.7 has to be fixed separately. At this carrier T-118 holds by Gelfand duality of the clock spectrum, — true, but assumed rather than obtained as a limit of O-clocks.
The finite depth register realizes an aperiodic history for any specified finite number of steps. Its scaling limit depends on how the mesh and the distances to both endpoints are chosen. Depth register
11.4 The depth register: an exact finite carrier
Proven [T] (Theorems 11.1–11.5; witnesses in website/scripts/check_core_numbers.py). The aperiodic parameter of T-53b has a finite carrier: a register that records the stratal depth (§10.3) as an ordered chain , not as a cycle. Relative to it the dissipative semigroup is the Page–Wootters conditional dynamics exactly at every reading, in a world of dimension , and the arrow holds on the whole recorded history; the full flow with is reproduced along each of its solutions; the continuum parameter is the scaling limit of the readings, with an explicit error, and its algebra is . Not obtained: a monotone functional on the whole orbit of any clock whose readings form a group orbit (Lemma of §11.3), and as a conditional law independent of the state (second Lemma of §11.3). Assumption, as for every Page–Wootters statement: that the world is in a timeless state of the kind constructed — the constraint half of A5 (T-87, step 4) — is not derived here; the theorems say that such a finite timeless world exists and what it yields.
Set-up. Fix . The system is a holon, . The environment is two further holons, , with a fixed reference vector . The depth register is with orthonormal readings ; the reading records that steps have occurred. It sits inside the O-registers of holons in positional notation, with : the lowest digit is the O-tick of §2–§3, and the higher digits count its windings — the quantity that relative to the O-clock alone is not an invariant observable (§11.2). Given unitaries on , the constraint is of Feynman–Kitaev type,
and the conditional state at reading is that of Definition 3.1, . The constraint is R. P. Feynman's clock ("Quantum mechanical computers", Found. Phys. 16, 507 (1986)), used by A. Yu. Kitaev (A. Yu. Kitaev, A. H. Shen, M. N. Vyalyi, Classical and Quantum Computation, AMS 2002, ch. 14), read as a discrete Page–Wootters constraint by J. R. McClean, J. A. Parkhill and A. Aspuru-Guzik ("Feynman's clock, a new variational principle, and parallel-in-time quantum dynamics", PNAS 110, E3901 (2013)) and studied as a system–time history state for a finite clock by A. Boette and R. Rossignoli ("History states of systems and operators", Phys. Rev. A 98, 032108 (2018)). What is added here is the environment and the dilation of step (i) below, which turn the unitary history into a dissipative one.
Let be any CPTP semigroup on , and . There are unitaries on , depending on , and but not on any state, such that:
(a) ; with , of dimension ; the eigenvalues of are , , each -fold, so the gap above the kernel is .
(b) For every state of the holon, , with and , satisfies , and
(c) If is unital and primitive (Theorem 10.1), then along the readings is non-increasing and the purity is strictly decreasing, unless .
Proof. (i) Dilations. For each the map is CPTP on ; its Choi matrix is , so it has Kraus operators (some possibly zero) with . The map is an isometry (W. F. Stinespring, Proc. Amer. Math. Soc. 6, 211 (1955)); completing its seven image vectors to an orthonormal basis gives a unitary with . Put and . Then and for every .
(ii) Kernel and spectrum. Since , conjugation by the unitary gives , where is the Laplacian of the path with vertices. Its quadratic form is , vanishing only on constant vectors, and its eigenvalues are . This is (a).
(iii) Conditional states. , of trace ; by (i) its normalised partial trace over is . With (ii), , so . This is (b).
(iv) Monotonicity. is CPTP and unital, and by the semigroup law. Monotonicity of relative entropy with gives the first claim. For the purity, the Kadison–Schwarz inequality for unital completely positive maps (M.-D. Choi, Illinois J. Math. 18, 565 (1974)) gives for Hermitian ; taking the trace, . Hence is non-increasing in . It is also real-analytic in , since the entries of are entire functions. If , then is constant on that interval, hence on ; primitivity gives , so and .
Why this does not contradict §11.2–§11.3. The almost-periodicity Lemma needs readings that form the orbit of a unitary group; here the step is a partial isometry, and no reading follows . The periodic-clock result of Chataignier, Höhn, Lock and Mele concerns clocks with a action, which the chain does not carry. The Unruh–Wald theorem concerns correlation with an external Schrödinger parameter; here there is none — the readings are the time. The price is visible: the recorded history has a first and a last reading. And relative to the lowest digit alone, the O-tick , the conditional state is the average of over : the O-clock still reads no arrow.
Close the chain into a cycle by adding the edge with . The state of Theorem 11.1 also satisfies the cyclic constraint and gives the same conditional states. Along the cycle every functional that decreases on the first steps increases on the wrap step. No periodic register does better: a function on a cycle cannot decrease on every step. For the seven readings of one O-clock the best possible is six decreasing steps of seven.
Proof. The holonomy around the cycle is , so conjugates the cyclic constraint to the cycle Laplacian , whose kernel is again spanned by ; the conditional states are computed as in Theorem 11.1. The rest is the fact that the increments of a function around a cycle sum to zero.
This is the finite form of the periodic-clock result (§11.2): the dynamics relative to a period- register is periodic, and the fraction of steps on which the arrow fails is , attained.
Let , , solve the full equation of §9.1 with , and a state, both continuous along the solution. Put and . Then:
(a) each is a Lindblad generator, and the propagators are CPTP with ;
(b) the construction of Theorem 11.1, with a dilation of , yields a finite timeless world whose conditional states are exactly at every reading .
The constraint of (b) depends on the solution; for a nonlinear no constraint independent of the solution does this (second Lemma of §11.3).
Proof. (a) with the replacement channel , and (since ) is CPTP for ; a sum of Lindblad generators is one. The time-ordered exponential of a continuous family of Lindblad generators is a limit of products of CPTP maps, and the CPTP maps form a closed set. Because , , so solves the linear equation ; uniqueness for linear equations with continuous coefficients gives . (b) Step (i) of the proof of Theorem 11.1 uses only that the maps it dilates are CPTP.
Every Lyapunov statement about the solution — the H-theorem of T-261, the monotonicities of Theorem 10.1 — therefore holds at the readings of the register, because the readings reproduce the solution.
In Theorem 11.1 let and . Then
for . For the full flow of Theorem 11.3 the same holds with . Hence a window is carried, with error at most in trace norm at every , by a finite world of dimension
that is, ; at the readings themselves the error is zero, and there is no recurrence inside the window.
Proof. with , and . For the bound on : , and . For the full flow , since .
Measure the readings in a macroscopic unit: , , where is the chronon in that unit and the reading taken as origin; the time algebra of the register is its diagonal algebra . Take a sequence of registers with , and . Then:
(a) for every , eventually every point of lies within of a reading: the reading sets converge to in the pointed Hausdorff sense;
(b) the sampling maps are -homomorphisms with for every ; hence is an injective isometric -homomorphism .
With the origin at the first reading () the same construction gives : the recorded time has a beginning, and is the limit seen from readings far from both ends.
Proof. (a) The readings form a grid of mesh covering , which eventually contains . (b) Evaluations are -homomorphisms, and . A function attains 's maximum at some ; by (a) some reading is within of eventually, so , where is the modulus of uniform continuity of . The norm on is .
The contrast with the other clocks of §3.8 and §11.3 is in the shape of the reading set: summed O-clocks fill a circle of fixed circumference, composite O-clocks with incommensurate frequencies fill a torus, the depth register fills a line. In the positional realisation, readings take holons, so the window is chronons; in a macroscopic unit the limit of Theorem 11.5 is reached with and .
Routes that were examined and not used. (1) Davies weak coupling (1974) gives only in the limit of vanishing coupling with an infinite environment, and its error constants for a finite environment are not explicit; the construction above is exact. (2) Quasi-ideal clocks (Woods–Silva–Oppenheim 2019) have readings that form a group orbit on a finite space, so by the almost-periodicity Lemma any dissipation relative to them holds only on a finite window, and only approximately; they are not needed. (3) Poincaré recurrence times of a finite environment depend on Diophantine properties of its spectrum and admit no computable bound of the form uniform in the Hamiltonian, so they cannot give a theorem with explicit constants. (4) An infinite register as the limit of the tower of strata: its Feynman–Kitaev constraint has no normalisable kernel (option (c) of §11.3), so it is not a carrier but only the limit of carriers. (5) Deriving the chain register from the axioms: not achieved. The register is built from holons, but that the world's timeless state is of the constructed kind is the constraint assumption of A5, as for the O-clock.
11.5 The de Sitter observer algebra and the holon tower
Since 2022 there is a precise sense in which an observer changes the algebra of a gravitating region. Quantum fields in a region have an algebra of type III₁, which has no trace and no finite entropy; adding an observer with a clock and imposing the gravitational constraint turns it into an algebra of type II, which has a trace, density matrices and entropies. In the static patch of de Sitter space the result is of type II₁: the trace is finite and there is a state of maximal entropy. Two points of contact with UHM follow. The observer is a clock — a Page–Wootters reference frame (§3, §11.1) — and the maximal-entropy state of the holon is its trace . This subsection lists what the literature establishes, then proves what can be proved about the relation (T-348), and says what does not follow.
Terms. A von Neumann algebra is an algebra of bounded operators closed under adjoints and weak limits; a factor is one whose centre is the scalars. Type I factors are full matrix algebras ; a type II₁ factor is infinite-dimensional and has a finite trace, normalised to ; type has a trace that is only semifinite; type III has no trace. A factor is hyperfinite (equivalently injective) if it is the weak closure of an increasing union of finite-dimensional algebras. The entropy relative to the trace of a state with density is ; in a II₁ factor with equality only for .
What the literature establishes.
- Chandrasekaran, Longo, Penington, Witten (CLPW). "An algebra of observables for de Sitter space", JHEP 02 (2023) 082, arXiv:2206.10780. Theorem within their model (, one observer on a geodesic). The matter algebra of the static patch is of type III₁ and its only operators invariant under the static-patch Hamiltonian are scalars. The observer has Hamiltonian on ("it is physically sensible to assume that the energy of the observer is non-negative"); the constraint is . Without the invariant algebra is the crossed product of by its modular flow, of type (Takesaki duality); the projection onto cuts it to a type II₁ factor with . The state of maximal entropy is the Bunch–Davies state times the thermal distribution of the observer's energy, and entropies equal the generalised entropy up to a state-independent constant. Replacing by keeps type II₁. Their §2.4 adds: a hyperfinite II₁ factor is isomorphic to the Murray–von Neumann algebra of infinitely many almost maximally mixed qubits; local algebras of quantum field theory "are believed to be always hyperfinite", so "we expect" to be that algebra. Their footnote 8: a clock whose energy is bounded below cannot tell time perfectly.
- Jensen, Sorce, Speranza (JSS). "Generalized entropy for general subregions in quantum gravity", JHEP 12 (2023) 020, arXiv:2306.01837. Argument at . A spatially compact region with an observer gives type II₁ and an entropy-maximising state; a region containing an asymptotic boundary gives type . Their examples include a ball in flat space with conformal matter: type II₁ is not tied to .
- Related constructions. S. Ali Ahmad, R. Jefferson, SciPost Phys. Core 7, 020 (2024), arXiv:2306.07323 (crossed products by the modular flow for subregions of general theories give type ); J. Kudler-Flam, S. Leutheusser, G. Satishchandran, Phys. Rev. D 111, 025013 (2025), arXiv:2309.15897 (for dynamical black holes the entropy of the dressed algebra is the generalised entropy); T. Faulkner, A. J. Speranza, arXiv:2405.00847 (the generalised second law from the crossed product); C.-H. Chen, G. Penington, arXiv:2406.02116 (slow-roll inflation and evaporating Schwarzschild–de Sitter give type factors, with no state of maximal entropy). Arguments within their models; none of them derives or a coupling constant.
- De Vuyst, Eccles, Höhn, Kirklin (DEHK). "Gravitational entropy is observer-dependent", arXiv:2405.00114. Theorems within the model. The observers of CLPW are quantum reference frames, and the CLPW description of states is the Page–Wootters one ("PW = CLPW"). The type and the entropy depend on the observer: a clock is ideal if its energy spectrum is the whole line, and then the algebra stays ; a non-ideal clock with energy bounded below gives II₁. A periodic clock (a harmonic oscillator) is an incomplete frame: its isotropy group is , and it resolves only properties of the field that are themselves periodic.
- Fewster, Janssen, Loveridge, Rejzner, Waldron. "Quantum reference frames, measurement schemes and the type of local algebras in quantum field theory", Commun. Math. Phys. 406, 19 (2025), arXiv:2403.11973. Theorem. If the field has a KMS state at some , the frame-invariant algebra has a semifinite trace; if the frame also has a KMS weight at the same temperature, the trace is finite; precise sufficient conditions for a type II₁ factor.
- Witten. "A background independent algebra in quantum gravity", JHEP 03 (2024) 077, arXiv:2308.03663. Proposal. The algebra along an observer's worldline, with the Hartle–Hawking state as universal state of maximal entropy; with only de Sitter vacua of different the definition gives sensible results. enters as a label of the vacua, not as an output.
- Operator-algebra results used below. F. J. Murray, J. von Neumann, Ann. Math. 44, 716 (1943): the approximately finite II₁ factor is unique. A. Connes, Ann. Math. 104, 73 (1976): every injective II₁ factor with separable predual is isomorphic to it, . U. Haagerup, Acta Math. 158, 95 (1987): uniqueness of the injective III₁ factor. H. Araki, E. J. Woods, Publ. RIMS 4, 51 (1968): types of infinite tensor products of matrix algebras (ITPFI). A. Connes, E. Størmer, J. Funct. Anal. 28, 187 (1978): in a III₁ factor any two faithful normal states are approximately unitarily equivalent. D. Buchholz, C. D'Antoni, K. Fredenhagen, Commun. Math. Phys. 111, 123 (1987): the split property makes local algebras hyperfinite. R. Verch, Lett. Math. Phys. 29, 297 (1993), and C. J. Fewster, Lett. Math. Phys. 105, 1633 (2015), arXiv:1501.02682: the split property for the free Klein–Gordon field and for locally covariant theories in curved spacetime.
Let be the algebra of the O-registers of holons — the carrier of the depth register of §11.4 — with the normalised trace , embedded by , (one holon added).
(a) The tower is . ; the union is the UHF algebra of type , and its closure in the representation of is a hyperfinite factor of type II₁, hence . For a state on , , with equality only at ; restricting to does not lower , and for a normal state of , . The number 7 survives only before closure: projections of have traces , never , while has projections of every trace in , and gives the same . If the added holon is read as the lowest digit of the depth register, reading marks the point of , and the trace of the projection onto the readings in differs from by at most : on the readings the trace becomes Lebesgue measure — the uniform weight of the history state of Theorem 11.1.
(b) CLPW is hyperfinite, under a named condition. If the matter net in the Bunch–Davies representation has the split property along an increasing sequence of regions exhausting the static patch — established for the free massive Klein–Gordon field (Verch 1993; Fewster 2015) — then , and are injective, and .
(c) What the isomorphism carries. Under (b) there is an isomorphism . It is not unique (two differ by an automorphism of ), and every such carries to : the maximal-entropy state of CLPW corresponds to the product state , and for every normal state. Nothing else is carried: no is singled out that maps to operators with a place or a time, and the observer's Hamiltonian, the modular flow of the Bunch–Davies state and the geometry of the patch are not transported.
(d) Only finitely many holons are alive. In every normal state of the reduced state of the -th holon tends to in trace norm, so only finitely many holons are viable (). If of them are viable, with nat, . An infinite product of one living state is not normal on : it generates an Araki–Woods factor of type III — when all eigenvalue ratios of are integer powers of one , III₁ otherwise. Type II₁ arises only for , whose purity lies below the window.
(e) A finite clock gives a type I algebra. Let be a factor on which generates a flow with a KMS state at and (the assumption of CLPW), and let the clock Hamiltonian have pure point spectrum — the O-clock , or the depth register at any finite . Then the part of invariant under is : type I, with no operator of the field in it.
(f) No and no . The type and the trace of are the same for every (every ) and every observer mass (); the tower gives the same for every factor dimension ; JSS obtain II₁ for compact regions in flat space. So no invariant of the identification fixes , its sign, or the regeneration rate , which (a)–(d) never use (cf. T-346). A finite truncation , at , gives — a strictly monotone reparametrisation of , not a derivation, because nothing in UHM fixes .
Proof. (a) . Every tracial state of is , so has exactly one tracial state; a tracial state whose representation had a non-trivial centre would split into two different traces, so the closure is a factor. The trace is faithful and normal on it, the factor is infinite-dimensional, so it is of type II₁, and as the closure of an increasing union of matrix algebras it is hyperfinite; Murray–von Neumann uniqueness gives . The density of relative to is , so . Monotonicity is that of relative entropy under the partial trace (G. Lindblad, Commun. Math. Phys. 40, 147 (1975)), and the limit is the martingale property of Araki's relative entropy along an increasing net with dense union (M. Ohya, D. Petz, Quantum Entropy and Its Use, Springer 1993, ch. 5). A projection of has trace , and is odd. The readings in number up to . The argument uses nowhere except in the arithmetic.
(b) By additivity . Split: between sits a type I factor , so ; the closure of an increasing union of injective algebras is injective (Buchholz–D'Antoni–Fredenhagen). A crossed product of an injective algebra by the amenable group is injective; a corner of an injective is injective (compose the expectation onto with ). is a II₁ factor (CLPW) on a separable Hilbert space; Connes' theorem gives .
(c) is a normalised normal trace on , and a II₁ factor has only one. CLPW show that their maximal-entropy state is , and entropies are defined through densities relative to the trace, which preserves. Two isomorphisms differ by an automorphism of ; the group of automorphisms moves every finite-dimensional subalgebra to other copies of , so the image of has no invariant meaning.
(d) Let . is dense in : pick , , . For in a factor the product trace gives , so , and . For the bound, take larger than every viable index: , with the total correlation. At fixed purity the entropy is maximal on one dominant eigenvalue plus equal others (P. Harremoës, F. Topsøe, IEEE Trans. Inf. Theory 47, 2944 (2001)), and this maximum falls as grows, so when . Two product states , are quasi-equivalent only if (D. Bures, Trans. Amer. Math. Soc. 135, 199 (1969)); for a living , (at fixed purity the fidelity with is largest on the same extremal spectrum, at ; checked numerically to ), so the product with diverges. The type is that of the Araki–Woods classification: for identical factors the asymptotic ratio set is the closed subgroup generated by the eigenvalue ratios.
(e) Write an invariant element as a block matrix over the eigenspaces of . Invariance means entrywise, where . For the entries lie in . For an eigen-operator has , so if it is a multiple of a unitary with ; the KMS condition gives , a contradiction. So the off-diagonal blocks vanish.
(f) (b)–(c) hold verbatim for every and every (CLPW §2.5), and (a) for every . The numbers are arithmetic; is strictly decreasing.
Status. (a), (d), (f) are theorems [T]. (b) and (c) are theorems under the split property, which is proved for free massive matter and assumed for interacting matter; (e) is a theorem under the assumption that CLPW themselves use. CLPW already expected the isomorphism of (c), with qubits in place of holons; what T-348 adds is the holon tower as the concrete approximating chain, the list of what an isomorphism can and cannot carry, the count of living holons (d), the verdict on finite clocks (e), and the no-go for and (f). Witnesses: check_core_numbers.py (holon_tower_trace_entropy_is_nonpositive_and_monotone, living_holon_costs_at_least_0344_nats_below_the_trace, lambda_as_a_holon_count_is_a_reparametrisation).
Physical reading [I]. Empty de Sitter space, the state of maximal entropy, corresponds to the tower in which every holon is at : the dead background, below the window. Excitations are normal states, and by (d) they carry finitely many living holons, each costing more than nat of entropy below the maximum; in CLPW a particle of energy at the centre of the patch costs . The comparison is an interpretation: (c) gives no map from holons to particles. The CLPW condition has one UHM counterpart, ; by (e) it is not enough, because the step from type III to type II needs a clock with continuous energy spectrum, and every finite clock of UHM has pure point spectrum; the limit of the depth register does have continuous spectrum bounded below (T-352, §11.6).
Open questions [Pr]. T-352 describes a bounded-below continuous energy spectrum for its specified operator. Whether the corresponding relational algebras converge to , and whether a physical scale selects the register level , remain open. T-352
11.6 Direction and the continuous clock
The foundation of UHM is an ∞-topos , read in homotopy type theory, where every identification is invertible: its objects are ∞-groupoids. The dynamics is not invertible: a dissipative channel has no inverse among channels, the depth of §10.3 grows, and T-348 left open a clock with continuous energy spectrum bounded below. This subsection asks what of the arrow a symmetric foundation can carry, where the direction lives, and whether UHM has the clock that T-348 lacked. It lists what the literature establishes, then proves T-352.
What the literature establishes.
- Directed type theory, synthetic ∞-categories. E. Riehl, M. Shulman, "A type theory for synthetic ∞-categories", Higher Structures 1, 147 (2017), arXiv:1705.07442. Theorems. Homotopy type theory with a directed interval and extension types; Segal types (unique composites up to homotopy), Rezk types (local univalence), covariant families and a dependent Yoneda lemma. Semantics: Reedy fibrant bisimplicial sets, where Segal and Rezk types are Segal spaces and complete Segal spaces. U. Buchholtz, J. Weinberger, "Synthetic fibered (∞,1)-category theory", Higher Structures 7 (2023), arXiv:2105.01724; J. Weinberger, "Strict stability of extension types", arXiv:2203.07194 (the system "has semantics in simplicial objects of an ∞-topos").
- Directed univalence. M. Z. Weaver, D. R. Licata, "A constructive model of directed univalence in bicubical sets", LICS 2020, 915–928. D. Gratzer, J. Weinberger, U. Buchholtz, "Directed univalence in simplicial homotopy type theory", arXiv:2407.09146 (theorem: a universe of discrete types whose homomorphisms are functions, built in triangulated type theory); "The Yoneda embedding in simplicial type theory", LICS 2025, 127–142, arXiv:2501.13229; "The ∞-category of ∞-categories in simplicial type theory", arXiv:2602.02218 (straightening–unstraightening inside the theory). E. Cavallo, E. Riehl, C. Sattler, "Directed univalence for simplicial objects in an ∞-topos", arXiv:2607.02420 (July 2026; theorem: directed univalence holds in simplicial objects of any ∞-topos).
- Other directed type theories. D. R. Licata, R. Harper, "2-dimensional directed type theory", MFPS 2011, ENTCS 276, 263 (2011); A. Nuyts, Towards a directed homotopy type theory based on 4 kinds of variance, MSc thesis, KU Leuven (2015); P. R. North, "Towards a directed homotopy type theory", MFPS 2019, ENTCS 347 (2019), arXiv:1807.10566 (a hom-type former interpreted in ); J. Neumann, T. Altenkirch, "Synthetic 1-categories in directed type theory", TYPES 2024, LIPIcs 336, 7, arXiv:2410.19520 (a directed-J rule, variance by modalities; a first step toward a directed higher observational type theory).
- Model-independent ∞-category theory and internal categories. E. Riehl, D. Verity, Elements of ∞-Category Theory, Cambridge 2022 (∞-cosmoi: one formal theory for quasi-categories, complete Segal spaces and their fibred and internal variants). L. Martini, "Yoneda's lemma for internal higher categories", arXiv:2103.17141, and L. Martini, S. Wolf, "Internal higher topos theory", arXiv:2303.06437: ∞-categories internal to an arbitrary ∞-topos are simplicial objects of it satisfying the Segal and completeness conditions.
- Directed spaces. M. Grandis, Directed Algebraic Topology, Cambridge 2009, arXiv:math/0111048 for part I (d-spaces, fundamental categories in place of fundamental groupoids, motivated by concurrency and by portions of space-time). Theorems within the framework; no directed analogue of a Grothendieck topos serves as a foundation there.
- Time operators and positive energy. W. Pauli, Handbuch der Physik 24/1 (1933), p. 140 (argument); in rigorous form: a self-adjoint and a projection-valued time observable covariant under form a system of imprimitivity of , and the Mackey–Stone–von Neumann theorem gives (P. Busch, M. Grabowski, P. J. Lahti, Operational Quantum Physics, Springer 1995, ch. III), so a Hamiltonian bounded below admits only a covariant positive operator valued time. R. E. A. C. Paley, N. Wiener, Fourier Transforms in the Complex Domain, AMS 1934 (functions supported on a half-line have Fourier transforms in the Hardy space ). H. Wold (1938) and B. Sz.-Nagy, C. Foias, Harmonic Analysis of Operators on Hilbert Space, North-Holland 1970 (isometries: the Wold decomposition and minimal unitary dilations). P. D. Lax, R. S. Phillips, Scattering Theory, Academic Press 1967 (translation on the line with outgoing subspace ).
What remains open in the literature. Directed type theory has models in simplicial objects of every ∞-topos, and directed univalence holds there (2026). It does not replace the ∞-topos: its types are interpreted inside one, as category objects. There is no established "directed topos" as a stand-alone foundation, and, beyond Grandis' motivating examples, no link in this literature between directed paths and physical time; Chapter 17 of the mathematics corpus names that link as a hypothesis [H].
Let be the readings of the depth register (§11.4) as a category, with one arrow exactly when ; in the limit. Let be the category whose objects are the states of a holon and whose arrows are CPTP maps.
(a) The arrow lives in a category, not in a groupoid. The maximal subgroupoid of is discrete ( objects, identities only); its ∞-groupoid completion is contractible (the nerve has Euler characteristic ). On an ∞-groupoid every function that does not increase along paths is constant on components. The history of Theorem 11.1 is a functor , , ; for unital and primitive no arrow , , is invertible in , and turns into a functor to the poset .
(b) Sharp readings with a first reading have no Hamiltonian. (i) On the step is an isometry with , purely non-unitary; is the closed unit disc and every is an eigenvalue of , with eigenvector ; no unitary on extends , and its minimal unitary dilation is the bilateral shift on , with spectrum the unit circle. (ii) In the continuum limit with the origin at the first reading (Theorem 11.5, ), the translations of are with on ; is closed and symmetric with deficiency indices , has no self-adjoint extension, and . (iii) On , no nonzero is supported in while its Fourier transform is supported in . So a clock whose sharp readings have a first reading and are shifted by its own evolution has no self-adjoint energy; its unitary dilation has energy spectrum — the ideal clock that keeps the algebra of type (DEHK) — and Pauli's theorem, in the rigorous form cited above, denies sharp covariant readings to an energy bounded below.
(c) The clock the constraint already contains. The dressed constraint of Theorem 11.1(a) is the path Laplacian . Its eigenvalues are exactly the quantiles of the arcsine law on , so the normalised counting measure of the spectrum is within of in Kolmogorov distance, attained. In the limit, on is bounded and self-adjoint; the cosine transform is a unitary carrying to multiplication by . Hence , purely absolutely continuous and simple, with density of states ; is not an eigenvalue. For every and every , the probability of a reading in tends to as and as .
(d) Type II₁ from the depth register. In the setting of CLPW — a type III₁ factor on whose modular flow is generated by , KMS at — take as the observer's clock with Hamiltonian , ( for a chronon ). The operators of invariant under form an algebra unitarily equivalent to , , with the observer energy of CLPW. It is a factor of type II₁; in the normalisation where CLPW's projection onto has trace , , and its state of maximal entropy is the Bunch–Davies state times the thermal law of the observer energy restricted to and renormalised. Under the split property of T-348(b) it is isomorphic to . At every finite the same construction gives type I (T-348(e)): the passage from type I to type II₁ happens exactly at , where the history state stops being normalisable (§11.3, option (c)).
(e) What is not fixed. The type and the trace structure in (d) are the same for every and every ; the only new scale is the bandwidth , and nothing in UHM fixes .
Proof. (a) An arrow of has an inverse only if exists too, that is . The ∞-groupoid completion of a category is the geometric realisation of its nerve, and a category with an initial object () has contractible nerve (D. Quillen, "Higher algebraic K-theory I", LNM 341, 85 (1973), §1). The non-degenerate -simplices of the nerve of are the chains of distinct readings, of them, and . On an ∞-groupoid every path has an inverse, so . Functoriality of is the semigroup law . A channel with a channel inverse is a unitary conjugation (standard; proved in full through Kraus representations as Lemma 17.1 of Chapter 17 of Part XVIII of the mathematics-foundations corpus), which preserves purity; by Theorem 11.1(c) lowers the purity of every , so it has no inverse. Monotonicity of relative entropy under channels (G. Lindblad, Commun. Math. Phys. 40, 147 (1975)) with gives the functor to .
(b) (i) and omits , so is not onto and no unitary on the same space agrees with it; strongly, so has no unitary part (Wold). , a vector of squared norm ; so the open disc lies in the spectrum of , hence in that of , and closes it. The minimal unitary dilation is Sz.-Nagy–Foias, ch. I. (ii) , and vanishes on , which forces on the domain. on ; has the solution , only , so the indices are and von Neumann's theorem excludes a self-adjoint extension. has the solution , square-integrable exactly when ; for such the range of is not dense, so ; for , and the range is dense, so . (iii) If vanishes on , its Fourier transform is the boundary value of a Hardy-space function (Paley–Wiener); a nonzero function cannot vanish on a set of positive measure of the boundary (F. and M. Riesz). The last sentence is (ii), the dilation of (i), and Pauli's theorem as cited above.
(c) has diagonal and on the off-diagonals. For : in the interior ; at the first site ; at the condition is , . , so . The functions , , are the eigenfunctions of on with , an orthonormal basis; so is unitary, and the first computation gives . This function is continuous and strictly increasing on from to ; the change of variable , , identifies with multiplication by on , absolutely continuous and of multiplicity one. For the escape, with , which tends to as (Riemann–Lebesgue lemma); sum over .
(d) By (c) there is a unitary with (the density of states changes only the unitary). carries the operators invariant under to those invariant under on . With on , . The invariant part of under is the crossed product , generated by and (CLPW §3; M. Takesaki, Acta Math. 131, 249 (1973)); belongs to it, and an operator is invariant exactly when it lies in . Since is of type III₁, the crossed product is a factor of type (Takesaki, loc. cit.), whose trace restricted to functions of is in CLPW's normalisation — the observer distribution of their maximal-entropy state. So , finite and nonzero; the corner of a factor by a nonzero finite projection is a factor of type II₁, and its normalised trace is the stated state. A corner of an injective algebra is injective, so T-348(b) and Connes' theorem give . The finite case is T-348(e), since has pure point spectrum; by (c).
(e) Nothing in (a)–(d) depends on the values of and beyond their positivity.
Numbers. The quantile identity and the Kolmogorov distance hold to for (: the windows of to holons), and the cosine vectors diagonalise with residual below . For and the start , the probability of a reading is at , at and at , and the mean reading grows as readings per unit of ( at ), while the weight beyond (the maximal velocity is ) is below at . The Euler characteristic of the nerve of is for . Witness: check_core_numbers.py (depth_register_constraint_is_a_continuous_clock_bounded_below, depth_readings_form_a_category_with_contractible_groupoid_completion).
Status. (a)–(c) and (e) are theorems [T]. (d) is a theorem in the model of CLPW (their , , , ) once the observer's clock is the depth register with Hamiltonian ; that identification is a premise, not a consequence of the axioms — as in §11.4, the constraint half of A5 is assumed, and here in addition that the constraint operator itself is the clock's energy. in (d) needs the split property, as in T-348(b).
What this changes.
- For the foundation. The arrow of T-53b and T-53c is a functor out of a category, , followed by a monotone functional; by (a) no ∞-groupoid built from the readings can carry it — in the maximal subgroupoid there is nothing to be monotone along, and the completion is a point. It needs no new foundation: Rezk's classifying diagrams of and of are complete Segal spaces (C. Rezk, Trans. Amer. Math. Soc. 353, 973 (2001)), which is where the Riehl–Shulman types live, and by Cavallo–Riehl–Sattler directed univalence holds in simplicial objects of any ∞-topos, so also over . The directed half that Chapter 17 of the mathematics corpus calls missing is thus available inside the symmetric foundation, as category objects; that the corpus should be rewritten in simplicial type theory is not claimed, and the identification of this direction with experienced time stays [H] there.
- For T-53b and T-118. No status changes. T-53b is the statement that is a functor and that decreases along it. Of the two limits of Theorem 11.5, is the directed one: by (b)(ii) its translations are isometries without a Hamiltonian; is its two-sided dilation, whose translation generator has spectrum .
- For T-348. Open question (1) is answered at the premise of (d): UHM has a clock with continuous spectrum bounded below, and it is not a choice of level spacing — it is the limit of the Feynman–Kitaev constraint that already carries the dynamics, with spectrum and the arcsine density. The type I verdict of T-348(e) for finite registers and the type II₁ of (d) are two sides of one fact: the constraint gains continuous spectrum exactly when its history state stops being normalisable. The difference from CLPW's clock is the upper edge: the observer energy is cut at , which changes the trace of the projection by and not the type.
What does not follow. The time reading of the continuous clock is not sharp in the energy picture: by (b)(iii) and Pauli's theorem no covariant sharp reading exists for , and the readings are not shifted by ; a packet spreads at up to readings per unit of , and the direction of the readings is the same for and — the unitary group does not know the arrow, the half-line of readings does. The theorem fixes neither nor , hence neither nor (T-348(f), T-346).
Open [Pr]. (1) Whether the relational algebras of the finite registers, of type I, converge to the type II₁ algebra of (d) in a sense that carries the trace (T-348, open (2)). (2) Whether anything in UHM fixes the chronon , the only scale in (d). (3) A formulation of the holon dynamics in simplicial type theory in which is a Rezk type over and T-53c becomes an internal statement.
Related documents:
- Axiom Ω⁷ — ∞-topos as unique primitive
- Dimension D (Dynamics) — connection to internal time
- Dimension O (Foundation) — role of internal clock
- Evolution Γ — equation with internal time
- Spacetime — emergent geometry
- Categorical formalism — ∞-groupoid and ∞-topos
- Critical purity — connection of P_crit to time
- Viability — temporal interpretation