10.  Recovering Quantum Mechanics

Quantum mechanics is the best tested theory in physics, so any deeper theory has to reproduce it exactly. This chapter is PST paying that bill: where the famous probability rule comes from, what actually happens during a measurement, and what the theory says about the experiments every physics student meets, from the double slit to Bell’s.

10.1.  Structural and specific predictions

A foundational theory makes predictions at two levels. Structural predictions follow from the form of the theory alone, independently of the specific content of any particular precausal configuration $C$. Specific predictions require knowledge of $C$ and concern the particular values of physical parameters in this universe. This chapter concerns structural predictions: consequences that hold for any instantiated geometry, regardless of the detailed tension distribution that produced it. These are the predictions that a future derivation of the Standard Model from PST would inherit as constraints; they are also the predictions through which PST can in principle be distinguished from competing frameworks experimentally.

A second point about the character of predictions is worth stating explicitly. PST is not primarily a replacement for existing physical theories at the level at which those theories are predictive. General Relativity and Quantum Mechanics already give correct answers to the questions they ask. PST operates at the level beneath those questions, at the level of why those theories work, why the spacetime arena exists, why matter inhabits it, and why the symmetries they assume hold. The predictions in this chapter are therefore predictions about the domain where existing theories break down, approach their limits, or require assumptions that PST renders derivable.

This chapter collects the structures of standard quantum mechanics that PST recovers rather than predicts anew: the Born rule, the measurement/collapse rule, and the canonical commutation relations are derived from the substrate machinery, confirming that PST reproduces known quantum mechanics rather than contradicting it. The genuinely distinguishing prediction, the short-distance $d^{-6}$ Casimir correction through which PST departs from standard physics and can be tested, has a chapter of its own (Chapter 15).

10.2.  The Born rule from the Bernoulli measure

A foundational question the projected field theory (§12.1) leaves to this chapter is the derivation of the Born rule: the identification of $|\psi|^{2}$ with measurement probability. This subsection provides a structural derivation rooted in the Bernoulli measure and the Hilbert-space structure established in Chapter 4.

The PST Hilbert space. The gradient structure derivation of Chapter 4 established that the space $L^{2}(\mathcal{P}(D),\mu)$ of square-integrable complex functionals on configuration space, with the Bernoulli measure $\mu=\bigotimes_{a}\mathrm{Bern}(1/2)$, is a separable Hilbert space. The inner product is

$$\langle\phi,\psi\rangle\;=\;\int_{\mathcal{P}(D)}\phi^{*}(C)\,\psi(C)\,\mathrm{d}\mu(C),$$ (93)

and the order parameter $\psi\in L^{2}(\mathcal{P}(D),\mu)$ is a unit vector (after normalisation). This is the quantum state space of PST, derived from the Bernoulli measure rather than postulated.

Physical observables as projectors. A physical observable $\hat{O}$ with outcomes $\{o_{i}\}$ and corresponding eigenspaces $\{\mathcal{H}_{i}\}$ defines an orthogonal decomposition $L^{2}(\mathcal{P}(D),\mu)=\bigoplus_{i}\mathcal{H}_{i}$. The corresponding projection operators $P_{i}:L^{2}\to\mathcal{H}_{i}$ satisfy $P_{i}P_{j}=\delta_{ij}P_{i}$ and $\sum_{i}P_{i}=\mathbf{1}$.

The $|\psi|^{2}$ probability from the Bernoulli measure. The Bernoulli measure $\mu$ assigns probability $\mu(S)$ to each measurable subset $S\subseteq\mathcal{P}(D)$. Near the threshold, the physical probability of obtaining outcome $o_{i}$ is the rate at which configurations in the eigenspace $\mathcal{H}_{i}$ cross the modal threshold $\tau$. The threshold-crossing rate for a configuration $C$ is proportional to the modal tension excess $|a_{\scriptscriptstyle LG}||\psi(C)|^{2}$ (from the linear stability analysis of Chapter 4): configurations where $\psi$ has large amplitude cross the threshold more frequently, and the crossing rate is proportional to $|\psi|^{2}$, not to $|\psi|$ or $|\psi|^{3}$. The probability weight associated with configuration $C$ crossing at outcome $o_{i}$ is therefore $|\psi(C)|^{2}\cdot\mathbf{1}_{\mathcal{H}_{i}}(C)$, giving:

$$P(\text{outcome }o_{i}\mid\psi)\;=\;\frac{\displaystyle\int_{C\in\mathcal{S}_{i}}|\psi(C)|^{2}\,\mathrm{d}\mu(C)}{\displaystyle\int_{\mathcal{P}(D)}|\psi(C)|^{2}\,\mathrm{d}\mu(C)}\;=\;\frac{\|P_{i}\psi\|^{2}}{\|\psi\|^{2}}\;=\;\|P_{i}\psi\|^{2}$$ (94)

for unit-normalised $\psi$. This is the Born rule.

Gleason’s theorem closes the argument. The probability assignment $P\mapsto\|P\psi\|^{2}$ satisfies all probability axioms and, crucially, it is additive on orthogonal projectors: $P(P_{i}\oplus P_{j})=\|P_{i}\psi\|^{2}+\|P_{j}\psi\|^{2}$ for orthogonal $P_{i}$, $P_{j}$. This additivity is guaranteed by the linearity of the Bernoulli integral: $\int_{S_{i}\cup S_{j}}|\psi|^{2}\,\mathrm{d}\mu=\int_{S_{i}}|\psi|^{2}\,\mathrm{d}\mu+\int_{S_{j}}|\psi|^{2}\,\mathrm{d}\mu$ for disjoint $S_{i}$, $S_{j}$. By Gleason’s theorem [54] applied to the separable Hilbert space $L^{2}(\mathcal{P}(D),\mu)$ (dimension $\geq 3$ for $|D|\geq 2$), every non-contextual probability measure on the orthogonal projectors has the form $P\mapsto\operatorname{Tr}(\rho P)$ for some density operator $\rho$ (Computation 135). That is a real constraint and not a vacuous one: it excludes every non-quantum probability assignment. But it does not by itself single out $P\mapsto\|P\psi\|^{2}$, and the difference matters here. The threshold-crossing mechanism above fixes the measure on the configuration algebra, that is on one maximal abelian resolution; and a density operator is not determined by its diagonal. For $|D|=2$ the pure state $\psi=\tfrac{1}{2}(1,1,1,1)$ and the maximally mixed $\rho=\tfrac{1}{4}\mathbb{1}$ assign identical probabilities to every configuration and differ by a factor of four on the projector onto $\psi$ itself.

What closes the gap is the pointer, not Gleason. A measurement of an arbitrary $\hat{O}$ terminates in a reading of an apparatus pointer, and the von Neumann coupling (96) carries $\psi\otimes|0\rangle\mapsto\sum_{i}c_{i}\,|o_{i}\rangle\otimes|\lambda o_{i}\rangle$, after which the pointer position is a configuration-diagonal quantity. Applying the configuration-basis rule to the coupled state returns $P(o_{i})=|c_{i}|^{2}=\|P_{i}\psi\|^{2}$ for arbitrary $\hat{O}$. So the Born rule is not an additional postulate of PST: it follows from the Bernoulli measure together with the von Neumann coupling and the unitary propagation between crossings, which §10.4 states plainly that PST assumes rather than derives. Gleason constrains the form; it does not supply the state.

10.2.1.  Two non-contextualities, and which one is assumed

The appeal to Gleason invites an objection worth answering directly, because the substrate is contextual by construction: P1 makes distinction the primitive, tension is defined by complement-asymmetry, and no elementary property carries a context-free value. How can a derivation that assumes non-contextuality sit inside an ontology of that kind?

The two conditions are distinct. Gleason’s hypothesis constrains the probability measure: the value assigned to a projector $P$ must not depend on which orthogonal resolution of the identity $P$ is taken to belong to, which is frame-function additivity in an arbitrary orthonormal basis (Computation 135). Within PST this is not an additional assumption but the measure-theoretic face of the substrate’s label-indifference, the same invariance that fixes $\mu$ uniquely (§2.5): a substrate with no favoured property cannot weight a projector differently according to the company it keeps. Kochen–Specker contextuality [193] constrains value assignments, that is, whether every observable can be given a definite value independently of what is measured alongside it. These are not the same condition, and they do not merely coexist. For $\dim\mathcal{H}\geq 3$ Gleason’s theorem admits no dispersion-free frame function and therefore entails that no non-contextual value assignment exists; standard quantum mechanics is the witness that the Born rule is Gleason-non-contextual while the theory is Kochen–Specker contextual.

The accounting is therefore favourable rather than awkward. What is assumed at the Gleason step is even-handedness of the measure, itself already forced by relabelling invariance; what is derived is the contextuality of values, which is the feature the substrate carries by definition. The apparatus-dependence of the modal threshold in §10.3, $\tau\mapsto\tau_{i}$, is the mechanism by which that derived contextuality is realised measurement by measurement.

The derivation above establishes the Born rule for measurements that decompose into orthogonal eigenspaces in $L^{2}(\mathcal{P}(D),\mu)$. The full measurement postulate (collapse of the wave function) follows from the threshold-crossing mechanism of P3 through the three closures and one condition of §10.3 (explicit von Neumann measurement [24] Hamiltonian, decoherence inheritance, threshold-crossing uniqueness + irreversibility theorem, and agreement with the Born rule). The fermionic-sector treatment at full functional-analytic rigour follows from the three closures of §11.3, two of them given A1 to A4 (convergence of the canonical anticommutation relations [CAR] to the continuum, Wightman axioms, substrate-derived Dirac equation). The canonical commutation relations hold in §10.6, given the identification of the projected Boolean gradient with $\partial_{\mu}$.

10.3.  The full measurement postulate from threshold-crossing

The Born rule of §10.2 delivers probabilities: the probability of measurement outcome $o_{i}$ given pre-measurement state $\psi$ is $\|P_{i}\psi\|^{2}$. The full measurement postulate adds a state-update rule: after observing outcome $o_{i}$, the post- measurement state is the projection $P_{i}\psi/\|P_{i}\psi\|$ (“collapse”). In standard quantum mechanics, both the probability rule and the state-update rule are postulated separately; in PST, the state-update rule emerges from the same substrate-threshold-crossing mechanism that delivers the Born rule.

10.3.1.  The threshold-crossing event as collapse

In the substrate-projection picture, a quantum measurement is the moment at which the substrate configuration $C$ crosses the modal threshold for an observable. Concretely, let $\hat{O}$ have spectral decomposition $\hat{O}=\sum_{i}o_{i}P_{i}$ on $\mathcal{H}=L^{2}(M,d^{4}x)$ and let $\mathcal{H}_{i}$ be the eigenspace of $o_{i}$. The substrate-projection of $\mathcal{H}_{i}$ under $\Pi$ corresponds to a subset $\mathcal{S}_{i}\subseteq\mathcal{P}(D)$ of substrate configurations whose projected wavefunction lies in $\mathcal{H}_{i}$, and for a configuration-diagonal $\hat{O}$ the collection $\{\mathcal{S}_{i}\}_{i}$ partitions $\mathcal{P}(D)$. That restriction is not decorative: a family of subsets induces multiplication-by-indicator projectors, so no two non-commuting observables can both partition $\mathcal{P}(D)$, which is why §10.4.3 finds the $\mu_{x}$ and $\mu_{z}$ partitions incompatible. For a general $\hat{O}$ the reduction to a partition runs through the pointer, as above.

When a measurement is performed, the substrate’s tension structure couples to the measuring device, modifying the modal threshold $\tau$ into an outcome-dependent threshold $\tau_{i}$. The substrate configuration $C$ crosses $\tau_{i}$ for some specific $i$, not for all $i$ simultaneously, because the threshold-crossing event of P3 is by construction a single irreversible directed transition (the parity-violating mechanism of Computations 8 and 11 underwrites the irreversibility). The selection of $i$ at threshold-crossing is therefore unique, and the post-measurement substrate is in a configuration $C^{\prime}\in\mathcal{S}_{i}$ projecting to a wavefunction in $\mathcal{H}_{i}$: this is the collapse.

10.3.2.  Probability of selection

The probability that the threshold-crossing event selects outcome $o_{i}$ is, by the Born rule of §10.2,

$$\mathbb{P}(\text{crossing for }o_{i}\mid\psi) =\mu\bigl{(}\mathcal{S}_{i}\cap\{C:|\psi(C)|^{2}\,\text{above threshold}\}\bigr{)}/\|\psi\|^{2}$$ (95)

$$=\|P_{i}\psi\|^{2}/\|\psi\|^{2},$$

which is the Born probability. The collapse and the Born probability are therefore not two separate postulates but two facets of the same substrate-threshold-crossing event: the event selects outcome $o_{i}$ with probability $\|P_{i}\psi\|^{2}$, and the post-event state is in $\mathcal{H}_{i}$ by construction.

The measurement postulate, both probability rule and state-update rule, is delivered structurally by the combination of (i) the configuration-basis rule on the Bernoulli measure, Gleason’s theorem fixing its form and the pointer coupling supplying the state (§10.2, giving the probability rule) and (ii) the threshold-crossing irreversibility of P3 (giving the state-update rule). No additional postulate is required beyond the unitary propagation between crossings, which PST assumes (§10.4).

10.3.2.1.  Closure (1): Explicit measurement-coupling Hamiltonian

The standard von Neumann measurement model couples system and apparatus via the impulsive Hamiltonian

$$H_{\mathrm{int}}(t)\;=\;\lambda\,g(t)\,\hat{O}\otimes\hat{P}_{\mathrm{app}},$$ (96)

where $\hat{O}=\sum_{i}o_{i}P_{i}$ is the system observable, $\hat{P}_{\mathrm{app}}$ is the apparatus pointer momentum, $g(t)$ is a short pulse of unit integral (idealised as $g(t)=\delta(t)$), and $\lambda\gg 1$ a strong coupling. Under $U=\exp(-i\int H_{\mathrm{int}}\,dt/\hbar)=\exp(-i\lambda\hat{O}\otimes\hat{P}_{\mathrm{app}}/\hbar)$, the joint evolution acts as

$$|\psi\rangle\otimes|0\rangle_{\mathrm{app}}\;=\;\Bigl{(}\sum_{i}c_{i}|o_{i}\rangle\Bigr{)}\otimes|0\rangle_{\mathrm{app}}\;\xrightarrow{\;U\;}\;\sum_{i}c_{i}\,|o_{i}\rangle\otimes|\lambda o_{i}\rangle_{\mathrm{app}},$$ (97)

where the apparatus pointer is displaced by $\lambda o_{i}$ when the system is in eigenstate $|o_{i}\rangle$. Substituting into the substrate-projection picture: the substrate configuration $C$ now feels a modified tension field through its coupling to the apparatus, with outcome-dependent threshold

$$\tau_{i}(C)\;=\;\tau_{0}(C)+\kappa\,\bigl{(}\lambda o_{i}\bigr{)}^{2},$$ (98)

where $\kappa$ is the substrate-apparatus coupling coefficient (a P3 parameter). The configuration $C$ crosses $\tau_{i}$ if and only if $|\Pi^{-1}(P_{i}\psi)|^{2}>\tau_{i}(C)$, that is, the substrate-projected weight on outcome $i$ exceeds the outcome-specific threshold. The outcome-dependent threshold $\tau_{i}$ of the abstract argument is therefore not a postulate but the structural consequence of the substrate’s response to the apparatus coupling (96).

10.3.2.2.  Closure (2): Decoherence inheritance

In macroscopic measuring devices, the apparatus is entangled with environmental degrees of freedom (the bath), and the apparatus-pointer states $|\lambda o_{i}\rangle_{\mathrm{app}}$ rapidly become quasi-orthogonal on the decoherence timescale $\tau_{D}\sim\hbar/(k_{B}T)$ for thermal bath temperature $T$. Environment-induced superselection (Zurek [106]) selects the pointer-eigenbasis $\{|o_{i}\rangle\}_{i}$ as the einselected basis on which off-diagonal density-matrix elements vanish. In PST: the substrate threshold field $\tau_{i}$ inherits this einselected basis without modification. The substrate configuration $C$ couples to the apparatus through the einselected pointer states, so the threshold-crossing event of P3 is sharply defined on the same basis as standard quantum measurement. PST makes no claim of independent decoherence theory; the standard einselection mechanism applies verbatim. This is the meaning of “decoherence inheritance:” PST extends standard quantum measurement by adding the substrate-projection layer, without rewriting the decoherence machinery.

It is worth being exact about what is being inherited, because the division of labour between the three closures depends on it and is easy to misread in PST’s disfavour. Decoherence does two things and conspicuously does not do a third. It selects the pointer basis, and it destroys interference in that basis. It does not make an outcome definite. The evolution is unitary, so the global state stays pure, and what is left on the system is an improper mixture: a reduced density matrix numerically identical to a proper statistical mixture, while the formalism carries no fact about which outcome obtained [206]. Computation 146 verifies both halves directly: global purity $1$ throughout, and agreement with a hand-built proper mixture to machine precision under every local observable tried. This is standard quantum mechanics, not a PST claim; it is the reason the measurement problem survives decoherence at all.

The three ingredients therefore do disjoint work. The configuration-basis rule on the Bernoulli measure, its form fixed by Gleason’s theorem, supplies the probabilities (§10.2); einselection supplies the basis and the loss of interference; and the threshold crossing of P3 supplies the definiteness, and with it a single world. The crossing is not a restatement of decoherence in substrate vocabulary. It performs precisely the step decoherence cannot, and it does so without a collapse postulate and without branching.

That also settles a question the projected phenomenology raises later. Computation 144 finds that no fringe-visibility measurement separates a world in which the crossing selects from one in which it does not, and it would be easy to read that as the crossing doing no work. The reading is backwards. Improper and proper mixtures are locally indistinguishable by construction (Computation 146), so no statistic gathered on the system could have separated them. The absence of a signature is what the formalism requires, not evidence of an idle layer.

10.3.2.3.  Closure (3): threshold-crossing uniqueness

We state and prove the substrate-level theorem that anchors the collapse rule.

10.3.2.4.  Theorem (threshold-crossing uniqueness)

Let $\hat{O}$ be a self-adjoint observable with spectral decomposition $\hat{O}=\sum_{i}o_{i}P_{i}$ on the projected Hilbert space $\mathcal{H}=L^{2}(M,d^{4}x)$. Under the measurement coupling (96) and the outcome-dependent threshold (98), the substrate threshold-crossing event of P3:

1. Selects exactly one outcome $o_{i}$: for the substrate configuration $C$, the set $\{i:|\Pi^{-1}(P_{i}\psi)|^{2}>\tau_{i}(C)\}$ is a singleton (or empty if no crossing occurred).

2. Is irreversible: the post-crossing configuration $C^{\prime}\in\mathcal{S}_{i}$ does not revert to $C\in\mathcal{P}(D)\setminus\mathcal{S}_{i}$ under the substrate’s directed tension dynamics of P2–P3.

Proof of (U).

By construction, the eigenspaces $\{\mathcal{H}_{i}\}_{i}$ are orthogonal: $\mathcal{H}_{i}\perp\mathcal{H}_{j}$ for $i\neq j$, so $|\Pi^{-1}(P_{i}\psi)|^{2}+|\Pi^{-1}(P_{j}\psi)|^{2}\leq\|\psi\|^{2}$ for $i\neq j$. Together with the strong-coupling condition $\kappa\lambda^{2}\gg\tau_{0}$, the outcome-dependent thresholds $\tau_{i}(C)=\tau_{0}(C)+\kappa(\lambda o_{i})^{2}$ are outcome-separated: for $i\neq j$ with $|o_{i}|\neq|o_{j}|$, the thresholds $\tau_{i},\tau_{j}$ differ by $\kappa\lambda^{2}(o_{i}^{2}-o_{j}^{2})$. In the strong-coupling limit, this gap exceeds any single $|\Pi^{-1}(P_{i}\psi)|^{2}$ that has not yet been concentrated on a specific outcome by the einselected basis (Closure 2 above). Therefore at most one $\tau_{i}$ is crossed by $|\Pi^{-1}(P_{i}\psi)|^{2}$, delivering uniqueness. In the degenerate case $|o_{i}|=|o_{j}|$ with $i\neq j$, the einselection of Closure (2) selects a pointer-basis projection that re-separates the thresholds before the crossing, so the same argument applies on the einselected subspace. ∎

Proof of (I).

The substrate’s asymmetric tension structure of P2 (the directed modal threshold of P3) is encoded by the parity-violation mechanism of Computations 8 and 11 (the $T(C)$- configuration breaks the substrate-side parity transformation of $\mathrm{SU}(2)_{L}$ irreversibly). Concretely, the threshold-crossing move $C\to C^{\prime}\in\mathcal{S}_{i}$ involves a directed flow on the substrate’s tension landscape: the configuration $C$ is at a threshold-saddle that flows to $C^{\prime}$ under the P2 asymmetric tension. The reverse flow $C^{\prime}\to C$ would require crossing the same threshold-saddle in the opposite direction, which requires a sign reversal of the P2 directed tension. The P2 asymmetric primitive, “modally distinguishable in a directed way”, forbids the sign reversal at the substrate level. Therefore $C^{\prime}\to C$ is forbidden in the substrate’s directed dynamics, and the threshold-crossing is irreversible.

The directed-tension argument generalises the parity-violation mechanism of Computations 8 and 11 to arbitrary observables: the $\mathrm{SU}(2)_{L}$ parity violation is the prototype example (“handedness” of the weak charge), but the same asymmetric P2 structure forbids the reversal of any threshold-crossing event. The measurement postulate’s irreversibility, standardly justified by appeal to thermodynamic entropy increase, is delivered in PST by the substrate-level asymmetric tension. ∎

Closures (1)-(3) above complete the measurement postulate derivation. The state-update rule (collapse) is a structural theorem from: (i) the von Neumann measurement coupling (96) which sets up outcome-dependent thresholds (98), (ii) the decoherence-einselection inheritance which sharpens the pointer basis, and (iii) the uniqueness + irreversibility theorem (U), (I) above which selects a single outcome. Together with the Born probability of equation (95), this is the full measurement postulate, both probability rule and state-update rule, structurally delivered by PST’s substrate machinery with no additional axiom beyond the assumed unitary propagation between crossings.

10.3.2.5.  Closure (4): agreement with the Born rule

Closures (1)–(3) settle the state-update rule. They leave a consistency question that should be stated rather than assumed, because the probability rule and the state-update rule are derived by different routes and have to agree. The Born rule comes from the configuration-basis rule on the Bernoulli measure (§10.2), Gleason’s theorem fixing its form, and depends only on the projected weights. The crossing rule, by contrast, selects an outcome by comparing those weights against a threshold. Nothing so far forces the two to give the same answer.

Write the question precisely. For a fixed state $\psi$ the weights $w_{i}=|\Pi^{-1}(P_{i}\psi)|^{2}$ are fixed numbers, so the run-to-run randomness is not in them. It is in the substrate configuration $C$, entering through the tension magnitude $|T(C)|$ that §4.1 tests against the threshold. Selection frequencies are therefore

$$\mathbb{P}(\text{outcome }i)\;=\;\mathbb{P}_{C\sim\mu}\bigl{(}\,C\text{ crosses }\tau_{i}\,\bigr{)},$$ (99)

and agreement with Born requires this to equal $w_{i}$. Computation 143 shows that it does under a condition it identifies and does not discharge: the tension variable must be uniform in its own quantile across configurations. Given that, $\mathbb{P}(\tau_{0}\text{ below the outcome-}i\text{ margin})=w_{i}$ exactly.

That condition is not available for free, and an argument that it was is withdrawn. The argument ran: every derivation downstream of (17) reads $T$ only through the ordering of the magnitudes, Computation 140 verifies invariance under monotone reparametrisation, the probability-integral transform is such a reparametrisation, so uniformity is a choice of scale rather than an assumption. The step that fails is the first. (98) does not compare $T$ against $T$; it compares $\tau_{0}(C)$ against the Born weights $w_{i}$, which come from the quantum state and carry their own normalisation $\sum_{i}w_{i}=1$. That is a cardinal read: it needs to know where $w_{i}$ sits on the tension scale, not merely how tension values order among themselves. Applying a monotone $\phi$ changes $\mathbb{P}(\text{outcome }i)$ from $F(w_{i})$ to $F(\phi^{-1}(w_{i}))$ for every $\phi\neq\mathrm{id}$, and the results are not even probability vectors: with $w=(0.5,0.3,0.2)$ and $\tau_{0}$ uniform, $\phi(x)=x^{2}$ gives $(0.71,0.55,0.45)$ summing to $1.70$, and $\phi(x)=x/2$, which is affine, gives $(1.00,0.60,0.40)$ summing to $2.00$.

Computation 140’s invariance looks applicable because it co-transforms both sides, setting $\tau^{\prime}=\phi(\tau)$ and checking $\mathrm{sign}(\phi(|T|)-\phi(\tau))$; that hypothesis is silently dropped here, where only one side transforms. Its own scope note says it audits the reads identified in the body, and this one was never among them. The dilemma is clean: if the crossing rule is a genuine read of $T$, then $T$ is not free up to monotone reparametrisation and uniformity is a substantive constraint on the substrate; if $T$ really is free, the rule fixes no frequencies at all. Either way Closure (4) does not close on its own, and the residue is a condition, which the rest of this subsection reduces to the non-degeneracy of $T$’s contingent content and P3’s typicality supplies (Appendix C).

What is established, and is worth having, is that the condition is satisfiable. The obstruction is ties. Configurations sharing a tension value cannot be separated by any order-preserving relabelling, so the largest such tie puts a floor on how uniform the variable can be made. Computation 145 measures it on the scalar exemplar $T_{\mathrm{ex}}$ of (17), by exact enumeration of all $2^{|D|}$ configurations. For generic distinction structures the floor is already small at $|D|=14$ and falls as $|D|^{-1.5}$, so it is negligible for any substrate large enough to carry an apparatus. For regular $(D,\delta)$ the floor is larger at those sizes, $T_{\mathrm{ex}}$ there depending on $C$ only through $|C|$.

A tempting shortcut fails (Computation 180). It would say: that is the same sub-class (17) already excludes, since $T_{\mathrm{ex}}$ fails complement-asymmetry on regular graphs, so the two exclusions coincide. Both halves of that are wrong. (16) is a condition on $v$, not on $\delta$: its genuine content is the non-balance $\bigl{|}\sum_{a}v(a)\bigr{|}>0$, as stated at §3.4, so it excludes no distinction graph whatever. And the regular graphs are not the failing class in any case, their floor vanishing as $|D|^{-0.5}$ on the exemplar $T_{\mathrm{ex}}$, slower than generic but vanishing; that exponent is an artefact of integrality and is corrected (Computation 187), the complete graph’s atom falling exponentially under a generic admissible $T$ rather than as a square root, so those families pass for a better reason than was recorded. What does fail is a different property: a bounded active part. A star has one centre and $K_{k}$ with isolated vertices has a fixed clique, so in both $T_{\mathrm{ex}}$ takes $\mathrm{O}(1)$ distinct values however large $|D|$ grows; the star keeps an atom of exactly $1/2$ at every $|D|$, half of all configurations omitting the centre and tying at zero, which floors the distance from uniform at $1/4$ under every monotone reparametrisation. Both satisfy (14) and (15), checked exhaustively, and (16) cannot touch them, so both are admissible. The consequence for the register: Born agreement is jointly a condition on the law of $T$ and on $\delta$, and the theory constrains $\delta$ not at all in the relevant respect, so discharging it looked to need a genericity constraint on the distinction graph rather than a calculation. Nothing downstream is touched, and the generic case Computation 145 establishes is confirmed rather than disputed.

The genericity constraint is not needed: the graph half is a theorem (Computation 190). What Computation 183 measured, that no reduced family keeps the forced atom finite, is proved in two lines. Any matching of $m$ pairwise disjoint distinctions gives $m$ independent events of probability $\tfrac{1}{4}$ under $\mu$, each the event that both ends of one distinction are present, and a distinction-free configuration avoids all of them, so the forced atom is at most $(3/4)^{m}$. Reducedness, P1(iii)’s consequence that no two properties share a distinction set (Computation 184), forces the matching large: the properties outside a maximal matching form an independent set, each has its distinction set inside the $2m$ matched properties, and injectivity bounds their number by $4^{m}$, so $|D|-2m\leq 4^{m}$. Together, for every reduced $(D,\delta)$ and every admissible $T$, the atom at zero is below $\max\bigl{(}(3/4)^{|D|/4},(|D|/2)^{-0.21}\bigr{)}$, a universal bound that vanishes in $|D|$ and is $10^{-17}$ at substrate scale. The bound is checked exactly against every family the earlier computations tabulated, and it is universal rather than sharp: the subset-graph family that Computation 183 built as the extremal reduced case has an atom that falls as $|D|^{-1}$ against the bound’s $|D|^{-0.21}$. With Computation 187, every atom but the forced one belongs to $T$, so what remained was a condition on the law of $T$ alone, that its contingent content does not tie. Computation 193 shows that condition to be one face of the non-degeneracy of the content, non-balance being the other, and Computation 195 that the maximum-entropy law on the content has a density, under which both faces hold almost surely. P3’s typicality is read as reaching the content of $T$ (§1.7), so the condition is supplied on the footing of A3 and has left the structural front; Appendix C carries it as a conditional theorem.

One scope limit, stated plainly. Equation (98) cannot deliver uniqueness and Born statistics at the same coupling. Raising $\kappa\lambda^{2}$ raises every threshold, which is exactly what drives the probability of two simultaneous crossings to zero and is what clause (U) rests on. But pushed far enough it also orders the outcomes by margin, and the survivor becomes the same one every run. Born statistics live at the weak-coupling end of that axis and uniqueness at the strong-coupling end (Computation 143). The measurement postulate as assembled here is therefore consistent in the regime where the einselected pointer basis of Closure (2) does the separating, with (98) supplying irreversibility rather than the selection statistics. Locating that regime sharply is open.

10.3.3.  Why this is not the many-worlds picture

Many-worlds interpretations of quantum mechanics also identify the measurement postulate with a branching of the wavefunction at the measurement event. PST differs in one essential respect: PST’s threshold-crossing of P3 is a single irreversible event delivering one configuration, by the directed-threshold mechanism that already structurally underwrites parity violation (Computations 8, 11). There is no branching into multiple co-existing post-measurement states; there is a single post-measurement state $C^{\prime}\in\mathcal{S}_{i}$ for the actually-realised outcome $i$. The PST collapse rule is therefore intrinsically one-world.

10.3.4.  Why this is not an objective-collapse theory

The rule above shares an aim with the objective-collapse programme. GRW and its continuous-spontaneous-localisation descendants [195, 196] also replace the measurement axiom with a dynamical law rather than an interpretation, and they are the closest existing relatives of the rule set out above. They differ from PST on the one point that is empirically live.

Those models add a spontaneous localisation rate to the Schrödinger evolution, scaling with particle number or mass, so that an isolated superposition localises on its own once it is large enough. PST has no such term, and its absence is structural rather than a choice of parameter. The threshold a configuration must cross is $\tau_{i}(C)=\tau_{0}(C)+\kappa(\lambda o_{i})^{2}$ of (98), in which $\lambda$ is the displacement of an apparatus pointer. For a system coupled to nothing, $\lambda$ has no value and the outcome-dependent thresholds $\tau_{i}$ do not exist; what remains is the unmodified modal threshold $\tau$ of P3, which is not outcome-dependent, so there is nothing to cross and no member of $\{\mathcal{S}_{i}\}_{i}$ to select. That is the argument already run for the undetected slit in §10.4, here stated in general: in PST collapse is triggered by coupling, not by size.

Three consequences follow, in increasing order of exposure.

• No size limit on coherent superposition. Nothing in the collapse machinery of this chapter depends on mass or particle number, so a superposition isolated well enough survives at any scale. Matter-wave interferometry has carried molecules beyond $25$ kDa [197], and each such result bounds the objective-collapse parameter space from one side while leaving PST where it was. This is a contrast rather than a test: the two accounts agree wherever the postulated collapse rate is too small to resolve.

• The falsifier is a collapse without a coupling. A localisation observed in a system coupled to no apparatus, at a rate rising with mass, refutes the rule of this chapter outright. No term in $\tau_{i}$ could produce it, and no parameter could be adjusted to accommodate it afterwards.

• PST cannot retreat by adding a rate, and this is where the exposure is sharpest. The one rate-like quantity in the theory is the equilibrium density $j_{\tau}$ of threshold-crossed configurations, and §16.5 defines it as a saturation density against the configuration-space measure of P3, the substrate having no spatial extent against which a proper-volume rate could be defined. That definitional choice is exactly what forces $w=-1$ rather than $w=0$: a per-emergent-volume creation rate gives Hoyle–Bondi–Gold steady state, not a cosmological constant. A spontaneous collapse rate is a per-volume, per-time quantity of precisely the excluded kind.

The third point is worth stating plainly, because it closes an escape route before the theory has need of one. Were spontaneous localisation observed, PST could not absorb it by fitting a rate the way a collapse model fits its localisation parameter; it would have to give up the pre-spatial framing of the substrate, and with it the derivation of $w=-1$ in §16.5. The collapse rule and the dark-energy equation of state are exposed together, and that is a sharper commitment than either looks on its own.

10.4.  The canonical experiments

The measurement postulate assembled in §10.3 has so far been exercised only on abstract observables. This section puts it to work on the experiments a reader arriving from physics tests a foundational proposal against first: the double slit, Wheeler’s delayed-choice arrangement together with the quantum eraser, and the Stern–Gerlach magnet. One caveat governs all three and is stated here once rather than three times. PST supplies no propagation rule of its own between threshold-crossing events. It derives the arena, the projected Hilbert space of §10.2 and the canonical commutation relations of §10.6, but it nowhere derives that the projected state evolves between crossings under a one-parameter unitary group. That evolution is assumed below exactly as it is assumed in §10.3. In consequence nothing in this section is a new prediction. The probabilities are the Born probabilities of §10.2, so every number PST returns for these experiments is the number standard quantum mechanics returns already, and no experiment of this family can distinguish the two. What is gained is an account of the measurement event itself, single, irreversible and one-world, with no observer in a privileged role and no branching. PST’s one falsifiable departure from standard physics remains the $d^{-6}$ Casimir correction of Chapter 15.

10.4.1.  The double slit and which-path information

The double slit poses two questions at once: why an interference pattern accumulates from individually localised detections [188] when the two paths are not distinguished, and why that pattern is absent when they are. Standard quantum mechanics answers the first by superposition and the second by the measurement postulate, and leaves the boundary between the two regimes to be drawn by hand, typically by an appeal to what counts as an observation. PST draws it structurally, and the criterion is the one already in force in P3: the fringes survive exactly when no threshold-crossing event occurs at the slits, and are destroyed exactly when one does.

Consider first the apparatus with no which-path detector. The substrate configuration $C$ passes the slit plane with nothing coupled to its tension structure there. No apparatus pointer momentum enters a coupling of the form (96), so the threshold in force is the unmodified $\tau_{0}(C)$ of (98), the modal threshold $\tau$ of §4.1, which is not outcome-dependent. With no outcome-dependent threshold there is no crossing; with no crossing there is no selection; and with no selection no member of the partition $\{\mathcal{S}_{i}\}_{i}$ has been singled out at the slits. The projected wavefunction arriving at the screen therefore retains the contributions of both slit regions. The first threshold-crossing of the experiment is the one at which the substrate couples to the screen, whose observable is position rather than path, and whose eigenspaces each receive both contributions. The Born rule of (94) evaluated on that state returns $|\psi|^{2}$ with both contributions present, which is the fringe pattern. Interference and pointlike detection are not in tension in this account. The first is a property of a state that has not crossed a threshold, the second is a property of the crossing event, which by P3 delivers one configuration and hence one localised mark; the pattern is the accumulated statistics of many such single crossings.

Now install a which-path detector at the slits. Its coupling is of the von Neumann form (96), with $\hat{O}$ the path observable and its two projectors $P_{1}$, $P_{2}$; the substrate’s tension structure feels the apparatus and the threshold becomes the outcome-dependent $\tau_{i}$ of (98). The uniqueness half of the threshold-crossing theorem of §10.3, clause (U), then applies at the slit plane: at most one $\tau_{i}$ is crossed, so exactly one path is selected. The einselected pointer basis of Closure (2) is what makes the two path outcomes separable in the degenerate case where the labelling assigns them equal magnitude. The irreversibility half, clause (I), underwritten by the directed parity-violating mechanism of Computations 8 and 11, fixes that selection: the post-crossing configuration $C^{\prime}\in\mathcal{S}_{i}$ does not revert. The state reaching the screen therefore carries the contribution of the selected slit alone, and the accumulated distribution is the sum of two single-slit distributions. Nothing has erased a pattern that formed earlier. The pattern never formed, because a configuration was already selected at the slits.

What this buys is that the distinction between measurement and no measurement is objective and physical. The discriminator is whether a coupling modifies $\tau$ into an outcome-dependent $\tau_{i}$, which is a question about the substrate’s tension structure and the device it is coupled to, and it has a determinate answer independent of anyone’s knowledge, attention or existence. A which-path detector whose record is never read destroys the pattern all the same, because the crossing occurred at the slits and the irreversibility of clause (I) does not wait upon a reader. An observer who watches the slits without coupling anything to $\tau$ removes no fringe. The anthropocentric residue that the measurement postulate carries in its standard presentation is therefore absent here, and so is the alternative of keeping every outcome. By the argument of §10.3 the crossing is single and the collapse intrinsically one-world, so the surviving contribution is the only one there is, and not one branch among co-existing others.

Partial which-path information sits between the two cases, and the limits of what the machinery settles there are worth stating plainly. The threshold separation on which the proof of clause (U) rests is $\kappa\lambda^{2}(o_{i}^{2}-o_{j}^{2})$, which falls away with the coupling strength $\lambda$, and that proof is given in the strong-coupling limit $\kappa\lambda^{2}\gg\tau_{0}$ only. A weak which-path coupling accordingly leaves the thresholds too little separated for a crossing to be forced at the slits on every passage.

It is tempting to read a prediction into that gap, and for a time this book did: if only some passages cross, the pattern should degrade rather than vanish, and the rate of degradation might be a signature PST could be tested by. Computation 144 closes that reading, and against it. Were a crossing at the slits a physical selection, only the non-crossing fraction would interfere, and the Englert–Greenberger–Yasin relation [203, 204] would be violated downward, most strongly at weak coupling, which is the regime in question. Atom-interferometric tests of that relation see no such deficit [205]. What the gap in clause (U) leaves open is therefore filled by the decoherence of Closure (2), which PST inherits verbatim rather than rederiving: the crossing is redundant with it, and the quantitative complementarity between fringe visibility and path distinguishability is the standard one, not something PST modifies (§19.9).

That is a narrower claim than the one this subsection might have made, and it is worth being explicit about what it costs. In the partial-coupling regime PST adds no new phenomenology; it reproduces standard quantum mechanics, as it must, since recovering quantum mechanics is what the projection is for. The claim that survives is structural rather than quantitative: the familiar phenomenology of the double slit follows from P1–P3 with no separate measurement axiom and no observer, and the line between the two regimes is drawn by the substrate itself rather than by a reader.

10.4.2.  Delayed choice and the quantum eraser

Wheeler’s delayed-choice arrangement [189] sharpens the measurement question by moving the experimenter’s decision late. A photon enters an interferometer, and only after it has entered is it settled whether the recombining element is in place, so that the two paths interfere, or absent, so that the path taken is recorded. The statistics follow the configuration finally chosen. The apparent difficulty is that the photon seems to have had to behave in one way or the other at the entrance, and that a choice made afterwards then fixes which. Stated in that form, the experiment appears to exhibit a later event acting upon an earlier one.

The premise is what fails, and the collapse rule of §10.3 says precisely where. A measurement in PST is the crossing of an outcome-dependent threshold $\tau_{i}$, and $\tau_{i}$ does not exist until the substrate’s tension structure couples to the measuring device. Before that coupling there is the modal threshold $\tau$ of P3 and nothing that refines it: the partition $\{\mathcal{S}_{i}\}_{i}$ of $\mathcal{P}(D)$ is defined by the eigenspaces of the observable $\hat{O}$, and $\hat{O}$ is whatever the apparatus, in the configuration it finally takes, measures. An interferometer with its recombining element in place and one without it do not measure the same observable; they induce different partitions of the configuration space, and (98) assigns them different thresholds. There is therefore no earlier assignment of $C$ to some $\mathcal{S}_{i}$ for the later choice to overturn. Exactly one threshold-crossing occurs in a run of the experiment, and it occurs at the coupling which the completed apparatus defines. The theory has nothing to revise because it posited nothing that stood in need of revision.

This must be separated from a stronger claim which PST does not make and does not need. Nothing in the foregoing permits a later event to act upon an earlier one. Retrocausation would require a causal relation running from the choice back to something already settled, and neither term is available here. There is no earlier settled fact. And the causal relation is not fundamental in this theory but is coinstantiated with the metric (Chapter 5), generated by the Lorentzian signature of the projection rather than presupposed by it. The substrate that carries $C$ possesses no temporal order of its own, its primitives being distinction and tension, neither of which orders anything. So “the choice came later” is a statement about the emergent order in which the laboratory record is written, and within that order it is entirely correct. What it is not is a statement which dates the substrate-side transition and places it before the choice. The paradox is generated by taking it to do both at once, and it is that conflation, rather than the order of the laboratory events, which PST declines.

The quantum eraser [190] is usually presented as the sharper case, on the ground that which-path information once recorded is afterwards destroyed and the interference returns. In PST the first half of that description cannot be right. A threshold-crossing is irreversible, by clause (I) of §10.3 and the directed-tension mechanism which Computations 8 and 11 underwrite: the post-crossing configuration $C^{\prime}\in\mathcal{S}_{i}$ does not return to $\mathcal{P}(D)\setminus\mathcal{S}_{i}$. Nothing an experimenter does afterwards un-crosses a threshold. What the erasing operation performs is a further measurement, upon a correlated partner system, whose own crossing is equally single and equally irreversible, and which thereby labels each already-recorded detection with a second outcome. Interference reappears when the record is sorted by that label [191]. It does not reappear in the record unsorted.

That distinction is not a technicality; it is the whole of the matter. The raw counts at the signal detector carry no fringes, and sorting cannot give them any, for the trivial reason that a partition of a record into subsets leaves the record itself untouched. Were it otherwise the unsorted counts would report the setting of a distant apparatus. The fringes live in the correlated subensembles, each picked out by a label which was written before any sorting was contemplated; selecting a subensemble alters no detection and creates none. The standing caveat of this section applies here with particular force, and is worth repeating in one line: nothing above is a prediction, and every number PST returns for a delayed-choice or eraser arrangement is the number standard quantum mechanics returns already. What is gained is not empirical content but an account of the measurement event itself, set in a framework where the temporal order is derived rather than assumed.

10.4.3.  Stern–Gerlach and the definite eigenvalue

The experiment first performed by Gerlach and Stern in 1922 [192] is the cheapest place to put the collapse rule to work, because the classical expectation and the observed result differ in the plainest possible way. A neutral atom carrying a magnetic moment of magnitude $\mu$ through an inhomogeneous field feels a transverse force $\partial_{z}(\mu_{z}B_{z})$, so its deflection is proportional to $\mu_{z}$. Classically the moment is a vector free to point anywhere on the sphere, $\mu_{z}$ takes every value in $[-\mu,+\mu]$, and a beam of randomly oriented atoms must paint a continuous band on the screen. It paints two spots. The discreteness is not itself a mystery for PST, because it is already present in the projected Hilbert space before any measurement is considered. The orthogonal decomposition into eigenspaces $\{\mathcal{H}_{i}\}_{i}$ over which the Born rule of §10.2 is stated has, for the moment of a spin carrier, two summands rather than a continuum. The Mosco limit delivers a four-dimensional Lorentzian $M=\mathbb{R}\times S^{3}$ whose local Clifford algebra is $\mathrm{Cl}(1,3)\cong M_{2}(\mathbb{H})$ (§13.7, Computation 66), and the substrate’s odd-parity sector projects to fields that are sections of the associated spinor bundle (§11.2). The two-valuedness of $\mu_{z}$ is a corollary of that module rather than an added assumption, although, as the closing paragraph of this subsection records, the present text nowhere draws the corollary explicitly.

The question PST is actually needed for is the second one: why a given atom arrives at one of the two spots rather than in some superposition of both. Here the inhomogeneous field is the tension-structure coupling. In the abstract argument the apparatus replaces the global modal threshold $\tau$ with an outcome-dependent $\tau_{i}$ through equation (98); in a Stern–Gerlach magnet that substitution is realised by an ordinary field gradient. The observable in the von Neumann coupling (96) is $\mu_{z}$ and the apparatus pointer is the atom’s transverse momentum, which the gradient displaces in proportion to it. PST adds no force to this: the deflection is the electromagnetism of the projected Standard Model Lagrangian (195) and nothing else. What PST adds is that once the coupling is on, the modal threshold is no longer one global number. It is worth noticing which branch of the uniqueness theorem the experiment falls into. The two eigenvalues are $\pm\mu$, so $o_{+}^{2}=o_{-}^{2}$ and the two thresholds of equation (98) coincide: uniqueness is therefore not delivered here by the strong-coupling threshold gap but by the degenerate clause of the proof of clause (U), in which the einselection of Closure (2) re-separates the thresholds on the pointer basis before the crossing. In the laboratory that clause is the screen. The magnet alone does not complete a measurement; the magnet correlates $\mu_{z}$ with a macroscopic position, and it is the irreversible registration of that position which fixes the basis the substrate threshold field then inherits.

Given the basis, the rest is the collapse rule applied verbatim. The configuration $C$ crosses one of the two thresholds and not both, by clause (U), and it does not cross back, by clause (I): the directed tension of P2 forbids the sign reversal that a reverse crossing would require. The atom therefore lands at one place on the screen, with probability $\|P_{\pm}\psi\|^{2}$ by equation (95), which for an unpolarised beam is one half on each spot. The emphasis belongs on the word one. The outcome that was not selected is not realised on some other screen or in some other branch; it is simply never instantiated, which is the content of the one-world reading argued in §10.3. A single irreversible directed threshold crossing has delivered a definite eigenvalue out of what the classical account expected to be a continuum, and it has done so without a second postulate beyond P3.

Three magnets in series, oriented $z$, then $x$, then $z$ again, make the shape of the account sharper, because the third magnet recovers both outcomes and a naive reading of irreversibility would forbid it. The first crossing leaves the substrate in a configuration $C^{\prime}\in\mathcal{S}^{(z)}_{+}$. The second magnet measures a different observable, whose eigenspaces induce a different partition $\{\mathcal{S}^{(x)}_{i}\}$ of $\mathcal{P}(D)$, and because $\mu_{x}$ and $\mu_{z}$ do not commute the two partitions are incompatible: no cell of one sits inside a cell of the other. Clause (I) says that the $z$ crossing cannot be undone. It does not say that the configuration carries a permanent $z$ label. What the first crossing fixed was membership of $\mathcal{S}^{(z)}_{+}$ and nothing further, and the $x$ magnet installs a fresh tension structure whose thresholds are separated on a basis bearing no relation to the first. The crossing at the second magnet is thus a new selection, weighted $|\langle x_{\pm}|z_{+}\rangle|^{2}=1/2$, and the third magnet finds both $z$ outcomes at equal weight. On the substrate account the re-randomisation is expected rather than anomalous, and for a structural reason rather than a statistical one: a threshold crossing records an eigenspace membership for the observable actually measured, and PST provides nothing else in which a record of it could survive an incompatible measurement.

What this subsection does not do should be stated as plainly as what it does. The standing caveat of the section applies here too, and in a concrete form: the atom has been treated as propagating coherently through the field region and between magnets, and no such propagation rule is derived anywhere in this work. Beyond it there are three further debts, all of them specific to this experiment. First, the two-valuedness itself is available but undrawn: $\mathrm{Cl}(1,3)\cong M_{2}(\mathbb{H})$ and the existence of the spinor bundle are derived under P1–P3, yet no passage in this book takes the step from the spinor module to the statement that a spin projection has exactly two eigenvalues. Second, nothing in P1–P3 has been shown to force the silver atom’s moment to be carried by a single unpaired electron; that is atomic shell structure, and it is imported. Third, the magnitude of the splitting is imported as well: the moment of a Dirac particle follows in principle from the minimally coupled Dirac term of equation (195), but the non-relativistic reduction that extracts it is performed nowhere in this work. The number of spots and the size of the splitting are therefore inherited rather than computed. What the subsection establishes is the narrower point: a definite eigenvalue, rather than a superposition, is what P1–P3 deliver at the screen, and it costs no measurement postulate and no branching.

10.5.  Composite systems, the reduced state and no-signalling

Everything above treats a single system. The state space is $L^{2}(\mathcal{P}(D),\mu)$, the observable is one self-adjoint $\hat{O}$ on it, and the threshold-crossing of P3 selects one of its eigenspaces. That is enough for the Born rule, for the collapse rule and for the three experiments of §10.4, none of which needs a notion of “two systems”. It is not enough for the correlation experiments, which are about two wings and about what one wing’s setting does to the other’s statistics. This section supplies what they need, and the supply is unusually cheap: the substrate’s measure is already a product measure, so the tensor structure is a theorem rather than an axiom.

10.5.1.  The bipartition is of the property domain, not of space

A composite system in standard quantum mechanics is two systems in two places, and the tensor product is written down to match. PST cannot start there, because at the substrate level there is no space to be in: $M=\mathbb{R}\times S^{3}$ is an output of instantiation (Chapter 5), and the only primitive available is the property domain $D$ with its distinction relation. What the substrate does offer is a way of splitting $D$ itself.

Let $D=D_{A}\sqcup D_{B}$ be a partition of the property domain into two disjoint non-empty parts. Nothing in P1–P3 privileges one such partition over another: $\delta$ is symmetric and irreflexive and treats all pairs alike, which is the same even-handedness that forced the Bernoulli measure in §2.5. A bipartition is therefore not a structure the substrate possesses but a division an experimenter imposes, by building two apparatuses that couple to disjoint sets of properties. “Subsystem” in PST means a part of the property domain, and that is the whole of what it means.

10.5.2.  Theorem (tensor factorisation over a bipartition)

Let $D=D_{A}\sqcup D_{B}$ and let $\mu=\bigotimes_{a\in D}\mathrm{Bern}(1/2)$ be the canonical measure of §2.5. Write $\mu_{A}=\bigotimes_{a\in D_{A}}\mathrm{Bern}(1/2)$ and $\mu_{B}$ likewise. Then the map

$$\beta\colon\mathcal{P}(D)\longrightarrow\mathcal{P}(D_{A})\times\mathcal{P}(D_{B}),\qquad\beta(C)=(C\cap D_{A},\;C\cap D_{B})$$ (100)

is a measure-space isomorphism carrying $\mu$ to $\mu_{A}\otimes\mu_{B}$, and it induces a unitary

$$L^{2}(\mathcal{P}(D),\mu)\;\cong\;L^{2}(\mathcal{P}(D_{A}),\mu_{A})\otimes L^{2}(\mathcal{P}(D_{B}),\mu_{B})\;=\;\mathcal{H}_{A}\otimes\mathcal{H}_{B}.$$ (101)

Proof.

That $\beta$ is a bijection is immediate: a subset of $D$ is determined by its trace on $D_{A}$ and its trace on $D_{B}$, and any pair of traces arises from exactly one subset, namely their union. Under the identification $\mathcal{P}(D)\cong\{0,1\}^{D}$ it is the canonical splitting of a product index set, $\{0,1\}^{D}=\{0,1\}^{D_{A}}\times\{0,1\}^{D_{B}}$.

For the measure, $\mu$ is by construction a product over the index set $D$, so it factors over any partition of that index set. On the cylinder sets of (9), which generate the $\sigma$-algebra, take finite $C_{0}\subseteq D$ and split it as $C_{0}=C_{0}^{A}\sqcup C_{0}^{B}$ with $C_{0}^{A}=C_{0}\cap D_{A}$:

$$\begin{split}\mu\bigl{(}\{C:C_{0}\subseteq C\}\bigr{)}&=2^{-|C_{0}|}\;=\;2^{-|C_{0}^{A}|}\cdot 2^{-|C_{0}^{B}|}\\ &=\mu_{A}\bigl{(}\{C_{A}:C_{0}^{A}\subseteq C_{A}\}\bigr{)}\,\mu_{B}\bigl{(}\{C_{B}:C_{0}^{B}\subseteq C_{B}\}\bigr{)}.\end{split}$$ (102)

The two measures agree on a generating $\pi$-system of finite total mass, so they agree (Dynkin), giving $\beta_{*}\mu=\mu_{A}\otimes\mu_{B}$.

The unitary is then the standard isomorphism $L^{2}(X\times Y,\mu_{A}\otimes\mu_{B})\cong L^{2}(X,\mu_{A})\otimes L^{2}(Y,\mu_{B})$ for $\sigma$-finite measure spaces, which both are, being probability spaces. Explicitly, $\psi\mapsto\psi\circ\beta^{-1}$ carries the inner product (93) to the product inner product, and the elementary tensors $\phi_{A}\otimes\phi_{B}\colon(C_{A},C_{B})\mapsto\phi_{A}(C_{A})\phi_{B}(C_{B})$ span a dense subspace. Computation 142 verifies the factorisation numerically at finite $|D|$, to machine precision. ∎

Two things about this deserve emphasis, because they are what make the rest work. The first is that nothing was added: the factorisation is a property of the measure the substrate already has, and that measure was forced by relabelling and complementation invariance, not chosen for convenience. Had $\mu$ been any non-product measure over $D$, (101) would fail and PST would have no composite systems at all. The second is that the factorisation holds for every bipartition simultaneously, which is the substrate-level statement of the fact that what counts as a subsystem is a choice of apparatus rather than a fact about the world.

10.5.3.  The reduced state and local observables

With (101) in hand the standard apparatus is available and means what it usually means. An observable is local to $A$ when its measurement coupling (96) involves only properties in $D_{A}$, so that it acts as $\hat{O}_{A}\otimes\mathbf{1}_{B}$. This is a definition, and it is the honest one: locality in PST is locality in the property domain, and it is prior to any spatial notion, which does not yet exist at the level where the definition is made.

For a state $\rho$ on $\mathcal{H}_{A}\otimes\mathcal{H}_{B}$ the reduced state on $A$ is the partial trace $\rho_{A}=\operatorname{Tr}_{B}\rho$, characterised by

$$\operatorname{Tr}\bigl{[}\rho\,(\hat{O}_{A}\otimes\mathbf{1}_{B})\bigr{]}\;=\;\operatorname{Tr}\bigl{[}\rho_{A}\,\hat{O}_{A}\bigr{]}\qquad\text{for every }\hat{O}_{A}\text{ on }\mathcal{H}_{A}.$$ (103)

By the Born rule of §10.2, which is derived rather than assumed, the statistics of every observable local to $A$ are therefore fixed by $\rho_{A}$ alone. A state is correlated across the bipartition when $\rho$ is not a product $\rho_{A}\otimes\rho_{B}$; this is a statement about the bipartition and says nothing about distance, there being none.

10.5.4.  No-signalling across the bipartition

Now the question Bell’s theorem makes urgent. Let the $B$ wing measure $\hat{O}_{B}$ with spectral projectors $\{Q_{j}\}_{j}$. By §10.3 the coupling replaces the global threshold $\tau$ with the outcome-dependent $\tau_{j}$ on the $B$ factor, and the threshold-crossing selects exactly one $j$, irreversibly. Over an ensemble of runs the outcome $j$ occurs with Born weight $\operatorname{Tr}[\rho\,(\mathbf{1}_{A}\otimes Q_{j})]$ and leaves the post-crossing state $(\mathbf{1}_{A}\otimes Q_{j})\rho(\mathbf{1}_{A}\otimes Q_{j})$ up to normalisation. The ensemble average over outcomes is therefore

$$\rho^{\prime}\;=\;\sum_{j}(\mathbf{1}_{A}\otimes Q_{j})\,\rho\,(\mathbf{1}_{A}\otimes Q_{j}),\qquad\operatorname{Tr}_{B}\rho^{\prime}\;=\;\operatorname{Tr}_{B}\rho\;=\;\rho_{A},$$ (104)

the second equality by cyclicity of the trace on the $B$ factor together with $\sum_{j}Q_{j}=\mathbf{1}_{B}$. The reduced state on $A$ is unchanged. Since by (103) every observable local to $A$ has its statistics fixed by $\rho_{A}$, no choice of $\hat{O}_{B}$ at the $B$ wing, and no outcome obtained there, alters any statistic available at $A$. This is no-signalling, derived rather than asserted, and its two ingredients are the tensor factorisation proved above and the Born rule of §10.2. Computation 142 checks it numerically on correlated substrate states, including states that violate the Clauser–Horne–Shimony–Holt (CHSH) inequality, and finds the $A$ marginal invariant to machine precision under every choice of $B$ setting.

The one-world reading of §10.3 is not weakened by this argument, but it does constrain what the argument says. In a single run the crossing at $B$ is single and irreversible and there is a fact about which $j$ occurred; (104) is a statement about the ensemble, as it is in standard quantum mechanics. What PST adds is that the ensemble is an ensemble of runs rather than a collection of co-existing branches, so the averaging in (104) is ordinary statistics over repetitions and not a sum over worlds.

10.5.5.  Bell correlations and emergent spatiality

Bell’s theorem [194] rules out local hidden-variable completions of quantum mechanics, where locality is a condition on how variables factorise across spatial separation. PST is not obliged to satisfy it in the form usually intended, and the reason can now be stated structurally rather than asserted. The substrate’s own notion of a subsystem is a part of the property domain, by the bipartition above; spatial separation is not a substrate notion at all, since $M=\mathbb{R}\times S^{3}$ is an output of instantiation (Chapter 5). Substrate-level structure therefore carries no antecedent obligation to factorise along distinctions that only exist downstream of it, and a correlated $\rho$ across a bipartition is not a puzzle but the generic case: product states are the special ones.

What must still hold, and now does hold as a theorem rather than as an expectation, is no-signalling: by (104) the marginal statistics at one part of the bipartition are independent of the setting at the other. PST’s position is thus the familiar one for any theory taking spacetime to be emergent, but with the load-bearing step supplied: the correlations are not mysterious because they were never spatial in origin, and no operational influence is transmitted because the reduced state is provably untouched.

One link in this chain deserves the pressure a critic would put on it, and it is narrower than it first appears. Everything above is exact at the level of the property domain: the factorisation, the reduced state and no-signalling all hold across a bipartition $D=D_{A}\sqcup D_{B}$. The experiments, however, are performed at two places, so the chain is only about experiments if a spatial separation delivers such a bipartition.

It does, and the reason is the construction of §4.10 rather than anything new. There $\rho:D\twoheadrightarrow M$ is a quotient map and a spacetime point is a fibre of it, an equivalence class of properties rather than a container they sit in. A region $U\subseteq M$ is therefore already a set of properties, namely $\rho^{-1}(U)=\bigcup_{p\in U}p$, and fibres of a function are disjoint, so disjoint regions give disjoint subsets of $D$. Setting $D_{A}:=\rho^{-1}(U_{A})$ and $D_{B}:=D\setminus D_{A}$ yields exactly the bipartition the theorem above needs, and it does so without any further assumption: $\rho$ is fixed by $(D,\delta)$ and the tension structure, not by the configuration, so $D_{A}$ is a subset of $D$ and not a configuration-dependent selection. The correspondence from places to parts of the property domain is thus immediate, and a reader should not be told the theory lacks it.

What is genuinely not established is narrower, and it is worth stating exactly, because it is what closing this line would have to address. Two things remain.

First, $\rho$ is not unique. The freedom to partition $D$ into fibres in more than one way consistent with the tension structure is PST’s diffeomorphism invariance (§4.10), so $D_{A}$ is defined relative to a choice of $\rho$. No $\rho$-independent statement can depend on that choice, and the metric relations that make the two wings spacelike-separated are preserved across admissible partitions by construction. But PST has not shown that “the $A$ apparatus” is a $\rho$-independent designation, and until it does, the invariance is an expectation rather than a theorem.

Second, PST has no account of an apparatus. Identifying the $A$ wing with $\rho^{-1}(U_{A})$ says that a laboratory device occupying a region is realised by exactly the properties whose fibres lie in that region. That is close to definitional given that a point is nothing over and above its property-bundle. Of the rest, less is missing than it appears: whether such a set is spatially compact is not an extra fact about the geometry, because (30) already fixes the distance as $d(\rho(a),\rho(b))=|T(\{a,b\})-\tau_{\mathrm{null}}|^{1/4}$, strictly monotone in the tension excess, so compactness in $M$ and tightness in $T$ are one condition stated twice (Computation 156). What PST genuinely does not derive is why $T$ has such clusters, which is a fact about the configuration rather than about the structure.

The honest summary is therefore this. PST derives composite systems, reduced states and no-signalling outright, and the passage from a spatial bipartition to a bipartition of $D$ is a corollary of the quotient construction rather than an assumption. PST never needs the naming of the $A$ apparatus to be $\rho$-independent, the factorisation theorem being universally quantified over bipartitions (Computation 151), and the compactness half reduces to (30). What it inherits is narrower than either: why the tension structure clusters at all. That is a question about what an apparatus is rather than about whether the geometry cooperates (§19.9.2).

10.6.  Canonical commutation relations from the Boolean gradient

The projected Hilbert space $\mathcal{H}=L^{2}(M,d^{4}x)$ carries a natural position operator:

$$(\hat{x}_{\mu}\,\psi)(x)\;=\;x_{\mu}\,\psi(x),$$ (105)

defined as multiplication by the spacetime coordinate $x_{\mu}$. The momentum operator is the image of the Boolean gradient under the projection $\Pi$:

$$\hat{p}_{\mu}\;=\;-i\hbar\;\Pi\!\bigl{(}\partial_{\mu}^{\mathrm{B}}\bigr{)},$$ (106)

where $\partial_{\mu}^{\mathrm{B}}$ is the Boolean (Efron–Stein) gradient (equation (20)) in the $\mu$-direction.

Continuum limit of the Boolean gradient. For a smooth function $f\in L^{2}(M)$ pulled back to $L^{2}(\mathcal{P}(D),\mu)$ via $\Pi$, the Boolean central difference acting on a single substrate element $a$ at position $x_{a}$ is:

$$(\partial_{a}^{\mathrm{B}}\psi)(C)\;=\;\psi(C\cup\{a\})-\psi(C\setminus\{a\})\;\xrightarrow{\;|dx_{a}|\to 0\;}\;2\,\partial_{x_{\mu}}f(x)\cdot\hat{e}_{\mu},$$ (107)

where $\hat{e}_{\mu}$ is the unit vector in the $\mu$-direction and the factor 2 arises because adding and removing $a$ displace the projected centre $x(C)$ by $+dx_{a}$ and $-dx_{a}$ respectively. Including the factor $\tfrac{1}{2}$ implicit in the projection gives $\Pi(\partial_{\mu}^{\mathrm{B}})=\partial_{\mu}$, so that $\hat{p}_{\mu}=-i\hbar\,\partial_{\mu}$ acts on $L^{2}(M,d^{4}x)$ as the standard momentum operator.

Derivation. Acting on any $\psi\in\mathcal{H}$:

$$[\hat{x}_{\mu},\,\hat{p}_{\nu}]\,\psi \;=\;-i\hbar\bigl{(}x_{\mu}\,\partial_{\nu}\psi\;-\;\partial_{\nu}(x_{\mu}\psi)\bigr{)}$$

$$\;=\;-i\hbar\bigl{(}x_{\mu}\,\partial_{\nu}\psi\;-\;\delta_{\mu\nu}\psi\;-\;x_{\mu}\,\partial_{\nu}\psi\bigr{)}$$

$$\;=\;i\hbar\,\delta_{\mu\nu}\,\psi.$$ (108)

The canonical commutation relations $[\hat{x}^{\mu},\hat{p}_{\nu}]=i\hbar\,\delta^{\mu}{}_{\nu}$ follow once $\hat{p}_{\mu}=-i\hbar\,\partial_{\mu}$ is identified on $L^{2}(M,d^{4}x)$. The product rule step (second line) is then trivial. The substantive step is the continuum limit (107): the identification of the Boolean gradient with the standard momentum operator. The heuristic argument above is made rigorous in the following paragraph via strong resolvent convergence.

Rigorous continuum limit via strong resolvent convergence. The Mosco-convergence result of §7.8 (equation (58)) makes the limit (107) rigorous under A1–A7 (equicoercivity A7 is a theorem given A1, A2, A4 and A3, its equidistribution condition A3 supplied by P3’s maximum-entropy sublimation, §7.1). Strong resolvent convergence of the rescaled Boolean Laplacian to $\Delta_{g}$ implies that for any $f\in H^{1}(M)$ and $\lambda>0$, the Boolean gradient $|D|^{-1/2}\nabla^{B}(\Pi^{\dagger}f)$ converges in $L^{2}(\mathcal{P}(D),\mu)$ to the Riemannian gradient $\nabla_{g}f$. Under $\Pi$, this maps $\partial_{a}^{B}\mapsto\partial_{\mu}$ in $L^{2}(M,\mathrm{d}^{4}x)$, so $\hat{p}_{\mu}=-i\hbar\,\partial_{\mu}$ is the strong limit of the projected Boolean gradient. The operator $-i\hbar\,\partial_{\mu}$ is essentially self-adjoint on $C_{c}^{\infty}(M)\subset L^{2}(M,\mathrm{d}^{4}x)$ by standard elliptic theory [99]; its closure is the unique self-adjoint extension on the Sobolev domain $H^{1}(M)$. The continuum-limit identification $\Pi(\partial_{\mu}^{B})=\partial_{\mu}$ is therefore not merely heuristic: the Boolean gradient converges to the standard momentum operator in the strong resolvent topology, and the resulting $\hat{p}_{\mu}$ is self-adjoint and densely defined. The canonical commutation relations are thereby established.

10.7.  Quantum fluctuations

Near the modal threshold, where $\varepsilon(C)=|T(C)|-\tau\rightarrow 0$, the order parameter $\psi$ is in the critical region: the two minima of the modal potential are about to separate but have not yet done so. The potential near $\psi=0$ is nearly flat in the direction of bifurcation, meaning that the order parameter requires almost no energy to shift significantly. Small fluctuations in the tension $\delta T$ produce large fluctuations in the instantiated metric:

$$\delta g_{\mu\nu}\;\sim\;\eta_{\mu\nu}\cdot\frac{\delta T}{2\,|T-\tau_{\mathrm{null}}|^{1/2}}\bigg{|}_{T\to\tau_{\mathrm{null}}}$$ (109)

This is the PST interpretation of quantum fluctuations. They are not a fundamental feature of the ontology, not irreducible randomness built into the fabric of reality. They are the geometric signature of tension configurations that are close to the modal threshold: configurations whose instantiation is marginal, whose order parameter has not settled firmly into one of the two degenerate minima.

The Heisenberg uncertainty principle, $\Delta x\,\Delta p\geq\hbar/2$, acquires a natural interpretation in this picture. Position is registered by the realisation map $\rho$, which assigns to each property a point in the instantiated manifold. Momentum is the projection of the directional tension gradient via $\Pi$. Near the threshold, $\rho$ is sensitive to small variations in $T(C)$ because the denominator in equation (30) is small near $\tau_{\mathrm{null}}$. This sensitivity means that determining position precisely, which amounts to resolving $\rho$ to high accuracy, requires probing configurations at separations close to $\tau_{\mathrm{null}}$, which in turn amplifies the uncertainty in the tension gradient that projects as momentum. The uncertainty relation is a lower bound on the product of these two projection sensitivities; it is a structural property of $\Pi$ near the threshold, not an empirical law imposed from outside.

Wave-particle duality has the same precausal interpretation. A configuration that is deep in the instantiated regime, far from the threshold with $\varepsilon(C)\gg 0$, has a sharply defined order parameter and projects through $\Pi$ as a localised density: a particle. A configuration hovering near the threshold has a diffuse, oscillating order parameter and projects as a spatially extended metric fluctuation: a wave. The transition between wave behaviour and particle behaviour is the settling of the order parameter from the near-threshold critical region into a stable minimum; when an apparatus is involved, that settling is the collapse of §10.3, the apparatus coupling shifting the threshold to an outcome-dependent $\tau_{i}$ and the configuration crossing it. Measurement in PST is therefore a physical interaction, but one that acts on the threshold rather than disturbing a pre-existing definite state: it is the process by which a near-threshold configuration is brought sufficiently far from $\tau$ that its projection becomes localised.

10.8.  Summary

This chapter recovers quantum mechanics rather than predicting beyond it. The Born rule follows from the Bernoulli measure and Gleason’s theorem on $L^{2}(\mathcal{P}(D),\mu)$ (§10.2); the full measurement postulate, state update included, follows from threshold-crossing, with the agreement condition on the law of $T$ stated where it enters (§10.3). The canonical experiments, the double slit, delayed choice and the quantum eraser, and Bell’s, come out as the postulate predicts (§10.4); composite systems carry the reduced state and no-signalling (§10.5); the canonical commutation relations are the image of the Boolean gradient under projection (§10.6); and quantum fluctuations are the critical region of the order parameter near threshold (§10.7). Measurement is a physical interaction on the threshold, not a disturbance of a pre-existing definite state.