7.  Energy and Matter as Projected Tension

Einstein’s equations say that matter tells space how to curve. What they do not say is where the matter came from; you have to put it in by hand. This chapter argues that in PST it was there all along, and that what is called matter and energy is the substrate’s leftover pull seen from inside.

7.1.  Results imported from the technical core

Everything from this chapter onwards leans on four analytic results; three are proved in the convergence chapter at the back of the book (§18.1 onwards) and the fourth, (C4), in §8.10. They are collected here, as statements rather than proofs, for one reason: a reader should be able to follow the physics of the chapters that follow without leaving them. Each result below is used later exactly as stated, so a later reference to the convergence chapter points at the proof of something already in hand, not at something the reader must go and learn.

(C1) The Mosco limit (§18.1). Under the seven assumptions A1–A7, the Boolean Dirichlet forms on $\mathcal{P}(D_{n})$ converge in the Mosco sense to the Laplace–Beltrami form on $M=\mathbb{R}\times S^{3}$, and the convergence is strong-resolvent, so the Laplace–Beltrami operator is identified as the generator of the limit. The equidistribution assumption A3 is not an extra hypothesis: P3 supplies it as a typicality property of maximum-entropy sublimation. The limit therefore closes under A1–A7 with A3 discharged by P3, and what it still rests on beyond P1–P3 is A1, which §18.1 carries as architectural input.

(C2) The $K$-uniform rate (§18.2, with the reduction lemma at §18.1.6). This hypothesis is proved for the i.i.d. surrogate, by a five-step Berry–Esseen argument with three closures, and in the flat-bulk case, the curved case following by bounded geometric factors in the Sobolev constants. From that rate, equicoercivity A7 follows by the reduction lemma; A7’s own condition A3 is supplied by P3 rather than assumed, so A7 is a theorem given A1, A2, A4 and A3. A1 remains architectural input, so this does not make the chain unconditional on P1–P3 alone. This is what keeps (C1) from resting on a hidden constant. The surrogate qualification is real and is carried forward wherever the rate hypothesis is used.

(C3) Dirac-operator convergence (§18.3). The spectral-triple data converge alongside the forms, which is what licenses treating $M$ as a noncommutative-geometric object downstream. This is the one of the four whose closure is partial: it is closed at the relaxed-criterion level, and the existence of a finite uniform bound over the operative range $k\leq k_{D}$ is an analytical theorem, but the sharp constant $3.7$ is derived at its maximiser and scanned elsewhere rather than proved over the whole regime. It is now less opaque than that sounds: the Dirac operator drops out of the problem entirely, the case $k=1$ closes in closed form, the large-$N$ limit of the deficit at fixed $k$ is in closed form, and the sharp value is an explicit algebraic number at the corner $(6,12)$ (§19.2.1, Computations 157, 213 and 214). Nothing below turns on the sharp value.

(C4) Kernel normalisation (§8.10). The kernel normalisation $\mathcal{N}$ appearing in Newton’s constant, $G=6\pi d_{0}^{2}/(c\,r_{0}^{2}\,\mathcal{N})$, is the trace $\mathcal{N}=\mathrm{Tr}_{\mathcal{H}_{F}}(\mathbf{1})$ over the finite internal factor, the fermionic multiplicity of one generation. It is therefore a finite number fixed by the matter content and not a free parameter, which is what makes the Einstein–Hilbert coefficient of Chapter 8 a derived quantity rather than a fitted one.

What these four buy, taken together, is the right to say that the projected geometry is $M=\mathbb{R}\times S^{3}$ with its Laplace-type operator, rather than merely resembling it. Every downstream statement that rests on the Mosco limit, the Einstein–Hilbert action from the Seeley–DeWitt expansion, the spectral-triple structure, and the Casimir prediction insofar as it uses the kernel, inherits that standing.

7.2.  The source problem in General Relativity

Einstein’s field equations cast the relationship between matter and geometry in the form of a source equation: the Einstein tensor $G_{\mu\nu}$, which encodes the curvature of spacetime, is set equal to $8\pi G$ times the stress-energy tensor $T_{\mu\nu}$, which encodes the distribution of matter and energy [1, 10]. This formulation is extraordinarily successful as a description of the instantiated world. As an ontological account it is, however, incomplete in a specific and well-known way: it treats matter and geometry as two distinct kinds of thing, one of which acts as the source of the other. The left-hand side of the field equations, geometry, is described as marble, smooth and exact. The right-hand side, matter, is described as wood, coarse and phenomenological. Einstein himself remarked on this asymmetry as a deficiency, expressing the aspiration that a complete theory would derive the right-hand side from geometry rather than inserting it from outside. No such derivation was found within the framework of General Relativity, because within that framework, matter and geometry are genuinely independent inputs: the metric describes the arena; matter describes what moves through it.

PST dissolves this distinction at the foundational level. Both geometry and matter derive from the same precausal source, the asymmetric tension functional $T(C)$, preserved across the phase transition $\Phi$ and projected by the tension projection operator $\Pi$. The geometry of the instantiated manifold is the structural expression of $T(C)$ in the domain of spatial relations; the matter content is the energetic expression of the same $T(C)$ in the domain of physical densities. They are not two inputs; they are two outputs of a single projection. The marble and the wood are cut from the same precausal stone.

7.3.  The stress-energy tensor as projected tension

In the isotropic, homogeneous approximation the precise statement is:

$$T^{(\mathrm{iso})}_{\mu\nu}(x)=\Pi(T(C))(x)\cdot g_{\mu\nu}(x)$$ (39)

where $\Pi(T(C))(x)$ is the projection of the precausal tension functional onto the point $x\in M$, as defined in equation (31). This equation should be read carefully. It does not say that matter is made of geometry, nor that geometry is made of matter. It says that both are made of something prior to either: the asymmetric tension of the precausal configuration, which the functor $\Phi$ converts into a Lorentzian manifold and which the operator $\Pi$ distributes as a scalar density across that manifold. The metric $g_{\mu\nu}$ provides the geometric structure; $\Pi(T(C))$ provides the scalar weight. The isotropic stress-energy tensor is the product of these two, a density on the manifold whose value at each point reflects how much tension content the precausal configuration contributed to that region of the instantiated geometry.

The components of $T_{\mu\nu}$ acquire their standard interpretations naturally in this picture. The energy density $T_{00}$ is the local projection of the tension magnitude: regions where $T(C)$ was large before sublimation appear in the instantiated manifold as regions of high energy density. The momentum flux $T_{0i}$ would reflect the directional distribution of tension across configurations that realise nearby points, and the pressure components $T_{ij}$ the isotropic or anisotropic character of the tension distribution; these directional components lie beyond the scalar weight that $\Pi$ supplies and require the tensor-valued projection $\Pi_{\mu\nu}$, constructed in the next section. None of these quantities is postulated: the scalar weight is derived from the single precausal input $T(C)$ through the action of $\Pi$, and the directional components follow from the same pairwise tensions once the dyad structure is in place.

7.4.  The tensor projection $\Pi_{\mu\nu}$

The scalar weight is the trace of something larger, and the larger object is built from material the theory already has. §32 fixed the scalar form’s scope: it is exactly the cosmological constant, not an approximation for isotropic sources. This section constructs the tensor that contains it, records the one completion that does not work, and says what stays open. Everything below is checked in Computation 155.

7.4.1.  Why the obvious completion fails

A stress tensor needs a direction per contribution, and PST’s tensions are pairwise, so the natural candidate is one dyad per pair: for $\{a,b\}$ with separation $\Delta^{\mu}=\rho(a)^{\mu}-\rho(b)^{\mu}$, take the unit vector $\hat{e}^{\mu}=\Delta^{\mu}/\sqrt{|\Delta\cdot\Delta|}$ and the dyad $\hat{e}_{\mu}\hat{e}_{\nu}$. That normalisation diverges as $\Delta\cdot\Delta\to 0$, which is precisely where $\tau_{\mathrm{null}}$ sits, since (30) puts the null condition at $T(\{a,b\})=\tau_{\mathrm{null}}$. The candidate is singular exactly on the surface the whole construction is built around. Measured: over clouds of $200$ to $800$ realisation points the per-pair value of the summed dyad swings by $56\%$ and does not settle, because a handful of near-null pairs dominate it.

That is worth recording rather than passing over. The light cone is not an awkward region of this construction; it is the construction’s own zero, and any completion that normalises by the interval will fail there.

7.4.2.  The separation dyad and its trace

Drop the normalisation, and weight the unnormalised dyad by a function of the pair’s own invariant:

$$\Pi_{\mu\nu}(x)\;=\;\tfrac{1}{2}\sum_{a\neq b}W\!\left(s_{ab}\right)\,\Delta_{\mu}\,\Delta_{\nu}\;B\!\left(x;\,\rho(a),\,\rho(b)\right),\qquad s_{ab}=\Delta\cdot\Delta,$$ (40)

with $B$ a bond kernel supported between the two realisation points. Nothing is normalised, so nothing is singular. And the trace is immediate: $g^{\mu\nu}\Delta_{\mu}\Delta_{\nu}=s_{ab}$, which (30) ties directly to $T(\{a,b\})-\tau_{\mathrm{null}}$. The scalar $\Pi$ of (31) is therefore not lost: it reappears as the trace of (40) rather than as a factor multiplying the metric. That is the structural correction the previous section’s result demands. A scalar times $g_{\mu\nu}$ can only be $\Lambda$; a scalar as a trace constrains nothing.

7.4.3.  What fixes the equation of state

If $W$ depends only on the invariant, then on an ensemble rotationally isotropic about a rest frame $u^{\mu}$ the sum (40) can only take the form

$$\Pi_{\mu\nu}\;=\;\rho\,u_{\mu}u_{\nu}\;+\;p\,h_{\mu\nu},\qquad h_{\mu\nu}=u_{\mu}u_{\nu}-g_{\mu\nu},$$ (41)

which is the perfect fluid. It is not imposed but forced by the ensemble’s symmetry, and the residual of the decomposition falls from $1.5\times 10^{-1}$ to $4.8\times 10^{-2}$ as the cloud grows from $200$ to $1600$ points.

The two coefficients then have a reading the theory already supplies. $\rho$ is the moment over timelike pairs and $p$ the moment over spacelike ones, and those two classes are separated by $T(\{a,b\})$ against $\tau_{\mathrm{null}}$, the same cut (30) makes to fix the Lorentzian signature. So the threshold that gives the light cone gives the equation of state, and $w=p/\rho$ is the balance between the tension carried on spacelike pairs and on timelike pairs. Nothing else enters.

Writing the weight as $+f(|s|)$ on timelike pairs and $-\lambda f(|s|)$ on spacelike ones makes $\lambda$ that budget ratio and the only free scalar. Then $w=-1$ is one scalar condition, $\rho+p=0$. Solving it in the profile family tested gives $\lambda^{*}=0.8699$, and there the measured $\Pi_{\mu\nu}$ is proportional to $g_{\mu\nu}$ to $3.7\times 10^{-2}$, which is exactly the decomposition residual, so at $\lambda^{*}$ the tensor is the scalar form. The cosmological constant of §16.5 is one point of a one-parameter family, recovered rather than assumed.

7.4.4.  The weight is not free either

The budget ratio above was written as though $W$ could be chosen. It cannot. Two objects the theory already has, made to agree, fix it completely (Computation 158).

(31) projects the tension itself, and the trace of (40) is $g^{\mu\nu}\Delta_{\mu}\Delta_{\nu}=s$. (30) ties the two together: $|s|=|T-\tau_{\mathrm{null}}|^{1/2}$ with $\mathrm{sgn}(s)=\mathrm{sgn}(\tau_{\mathrm{null}}-T)$, so

$$T(\{a,b\})-\tau_{\mathrm{null}}\;=\;-\,s\,|s|$$ (42)

identically, for timelike, null and spacelike pairs alike. Requiring the tensor projection to reproduce the scalar one on its trace, pair by pair, is then one equation in one unknown,

$$W(s)\,s\;=\;T(\{a,b\})-\tau_{\mathrm{null}}\;=\;-\,s\,|s|\qquad\Longrightarrow\qquad W(s)\;=\;-\,|s|,$$ (43)

with no free function and no free scalar beyond the $8\pi G$ unit conversion (39) already carries.

One thing must be added, and nothing is read off it. With (43) and no range restriction the long pairs dominate and the answer drifts with the size of the region sampled, which a density must not do. PST supplies exactly one candidate scale for the restriction, the coherence length $d_{0}$, imposed on the invariant at $|s|\lesssim d_{0}^{2}$; once the window exceeds it the answer converges, to the same value for four different cutoff shapes and for cutoff scales differing by a factor of four (Computation 158). The cutoff is needed for convergence; no number comes out of it.

It is not a range restriction, and calling it one would mislead. The set $|s|\lesssim d_{0}^{2}$ is a neighbourhood of the light cone, which is not compact: a pair separated by a great distance still satisfies it, provided the separation is nearly null. Measured on a scattered cloud in a box of half-width $40$, $98\%$ of the surviving pairs are separated spatially by more than seven correlation lengths, with a median of $31$ (Computation 159). That is inert for the vacuum and for radiation, and decisive for everything else, which is why the next subsection exists.

What that forces is radiation. By construction the trace of $\Pi_{\mu\nu}$ is the sum of $T(\{a,b\})-\tau_{\mathrm{null}}$ over pairs, and that vanishes in the large-window limit: the null threshold is the balance point of the pair-tension distribution. So the two coefficients of (41) satisfy $\rho=3p$, and

$$w\;=\;\tfrac{1}{3}$$ (44)

exactly, with $(\rho-3p)/\rho$ shrinking from $-1.9\times 10^{-1}$ to $-5\times 10^{-6}$ as the window grows. It is also independent of the configuration: uniform, radially graded, Gaussian in space, Gaussian in space and time, and clustered into ten clumps all return $w$ between $0.3333$ and $0.3338$. The pair-tension projection is conformal matter, and that is forced rather than chosen, which is what a pair structure carrying no scale of its own ought to give.

7.4.5.  What is not derived

Two items were carried here. Computation 162 and Computation 163 have moved both, and what is left over is a third thing that was already named elsewhere.

First, conservation, and it now has two answers rather than none. $\nabla^{\mu}\Pi_{\mu\nu}=0$ is what the field equations require, and the direct route to it does not close. Read covariantly a pair has no separation vector; the stand-in is Synge’s world function derivative $\sigma_{\mu}$, with $s=\sigma^{\mu}\sigma_{\mu}=2\sigma$. Two identities are exact, that one and its gradient $\sigma^{\mu}\sigma_{\mu\nu}=\sigma_{\nu}$, and together they collapse the per-pair divergence to

$$\nabla^{\mu}\bigl{[}f(\sigma)\,\sigma_{\mu}\sigma_{\nu}\bigr{]}\;=\;\bigl{[}2\sigma f^{\prime}(\sigma)\;+\;f(\sigma)\,\Box\sigma\;+\;f(\sigma)\bigr{]}\,\sigma_{\nu},$$ (45)

so all the curvature dependence sits in one scalar, $\Box\sigma$, whose departure from its flat value $4$ is $-\tfrac{1}{3}R_{\alpha\beta}\sigma^{\alpha}\sigma^{\beta}$ to leading order. On constant curvature that is exact in closed form, $\Box\sigma=1+3(d/L)\cot(d/L)$, and with the derived weight (43) the relative shift is $-\tfrac{1}{7}(d/L)^{2}$. Since the pair sum is cut at $|s|\lesssim d_{0}^{2}$, the separation obeys $d\lesssim d_{0}=2.2$ am and the violation is minute everywhere the theory is used: $6\times 10^{-43}$ at a solar-mass horizon, $6\times 10^{-45}$ at a neutron star’s surface, $8\times 10^{-59}$ at the Earth’s, $5\times 10^{-88}$ at the cosmological curvature scale (on the data’s bound $d_{0}\leq 2.3$ nm each is eighteen orders larger and still minute; the four values are Computation 162’s table re-evaluated at the derived $d_{0}$). That is the same parameter as (66), where $(kd_{0})^{2}=1.9\times 10^{-42}$ decides whether the projected equations reach a horizon at all, so conservation can never be the dominant error. That much is unconditional.

And the exact answer, which costs an assumption rather than an approximation. §7.5 identifies $\Pi_{\mu\nu}$ as the conformal part of $T^{\mathrm{(fields)}}_{\mu\nu}$, and diffeomorphism invariance of $S_{\mathrm{matter}}$ makes that tensor conserved on shell. Splitting it as $T_{\mu\nu}=\Pi_{\mu\nu}+\tfrac{1}{4}g_{\mu\nu}T$ with $\Pi$ traceless gives one line,

$$\nabla^{\mu}\Pi_{\mu\nu}\;=\;-\tfrac{1}{4}\,\nabla_{\nu}T,$$ (46)

exactly, on any background: the traceless part of a conserved tensor is itself conserved precisely when the trace is covariantly constant. Computation 158 showed the trace of $\Pi_{\mu\nu}$ vanishes, for every configuration tried, because the null threshold is the balance point of the pair-tension distribution. So the right-hand side is zero identically and $\Pi_{\mu\nu}$ is conserved exactly, with no $(d_{0}/L)^{2}$ residue at all, in the sector it actually describes. Computation 163 verifies (46) on a Friedmann background with a massive scalar, where the trace is not constant and both sides are nonzero, and confirms that the electromagnetic stress tensor is traceless for an arbitrary field strength, as the worked instance of a source in the class where the right-hand side vanishes.

So the construction does not need an action of its own, which is worth saying because it was carried here as though it did. What it needs is the identification, which Computation 199 leaves without obstruction (below), not the conservation.

Second, the divergence-free ambiguity, which is smaller than its name. A stress tensor assembled from pair contributions is fixed only up to the double divergence of a tensor with the algebraic symmetries of Riemann, the standard Irving–Kirkwood and Hardy freedom. For the one-scalar family $H_{\mu\alpha\nu\beta}=f\,(g_{\mu\nu}g_{\alpha\beta}-g_{\mu\beta}g_{\alpha\nu})$ the shift is $\Delta_{\mu\nu}=g_{\mu\nu}\Box f-\nabla_{\mu}\nabla_{\nu}f$, and Computation 163 computes three things about it exactly: it is symmetric, so it is an admissible regauging; its trace is $3\Box f$; and its divergence is $-R_{\nu\sigma}\nabla^{\sigma}f$, which is identically zero in flat space and curvature-suppressed otherwise, the same shape as the first item.

And tracelessness closes most of it. Keeping $\Pi_{\mu\nu}$ traceless is not a convention here: Computation 158 forces it on every balanced cloud, and out of balance the identification concerns the conformal part of the sum, which is traceless by definition (Computation 199). By the trace above that restricts the gauge to $\Box f=0$, and a harmonic gauge function that is localised is identically zero, by uniqueness for the Dirichlet problem on the Riemannian section and by the domain of dependence in Lorentzian signature, both checked numerically. So no regauging confined to a region exists at all, and what survives is a global harmonic function, which changes no integrated quantity. The tensor is determined up to that, which is a long way from being a representative of an arbitrary class.

What is left, then, is the conformal identification, and it is no longer open. Both results above lean on $\Pi_{\mu\nu}$ being the conformal part of $T^{\mathrm{(fields)}}_{\mu\nu}$, and §7.5 shows that nothing obstructs it: the conformal part of the pair sum reaches every metric-traceless tensor that obeys the null energy condition (Computation 199), and its magnitude is the state’s (Computation 196). The item was carried in three places, then in one, and closed there.

One correction comes with all this, because the earlier wording invited the wrong reading. With the derived weight the per-pair divergence is not zero even in flat space; it is $-14\sigma\,\sigma_{\nu}$. “Checked in flat space” was therefore never a statement about $W$. It is a statement about the sum over pairs and the bond kernel, and Computation 162 measures it as such: the residual falls with the pair count, at a rate consistent with a finite-sample effect rather than a systematic.

The cosmological constant was never among these, and its absence from this sector is by design rather than by default. Computation 155 showed that $w=-1$ is exactly the scalar form $\Pi\,g_{\mu\nu}$, which §16.5 derives separately as the vacuum term. The two sectors do different jobs and add: $T_{\mu\nu}=\Lambda\,g_{\mu\nu}+\Pi_{\mu\nu}$, the first with $w=-1$ and the second traceless.

7.4.6.  What the matter sector can and cannot be

Real matter is not a scattered cloud of realisation points; it is made of worldlines. Separating the pairs that lie along one particle’s own worldline from the pairs between different particles turns the equation of state into something recognisable, and then rules something out.

The worldline sector is exactly kinetic. Taking only the intra-worldline pairs, the projection gives

$$w\;=\;\frac{\beta^{2}}{3},$$ (47)

exactly, verified to $5\times 10^{-15}$ over nine speeds (Computation 159). At $\beta=0$ the pressure is zero to every printed digit, which is dust; at $\beta\to 1$ it is $1/3$. Equation (47) is the textbook kinetic equation of state of a gas, and nothing was put in to obtain it: it follows from the weight (43), which was itself forced.

A photon contributes nothing from its own worldline, and that is why radiation is $1/3$. A massless particle’s worldline is null, so the invariant along it is $s=0$ exactly, and the derived weight $W(s)=-|s|$ vanishes identically on it. Measured, the summed weight over the intra-worldline pairs of null worldlines sits fourteen orders below the same configuration at $\beta=0$, consistent with round-off in $s=\Delta\cdot\Delta$. So the radiation value is not a statistical limit or a high-temperature approximation. It is forced by the weight vanishing on null separations, and the separation that is null is the one $\tau_{\mathrm{null}}$ defines: the same cut that fixes the signature and gives the light cone. The photon enters the stress tensor exactly where it enters the geometry, through the same object.

But the worldlines are a vanishing share, and that closes the matter question. Writing $f$ for the worldline share of the energy density, the two sectors mix as $w=\bigl{[}1-f(1-\beta^{2})\bigr{]}/3$, so dust needs $f\to 1$ as well as $\beta\to 0$. It does not get there. Because $|s|\lesssim d_{0}^{2}$ selects a neighbourhood of the light cone rather than a compact region, the near-null background never dilutes: sweeping the density over fifty times the correlation length leaves $f$ between $0.01$ and $0.05$ throughout. The bond kernel of (40) is the obvious remaining candidate and it cannot help either, because it is normalised: $\int B\,\mathrm{d}^{4}x=1$, so the volume-averaged density is identically the integrated sum, whatever $B$ is. It decides where the stress sits, not what the equation of state is.

So the projection is strictly conformal for homogeneous matter, and dust is excluded rather than merely unfound. The rest mass of ordinary matter does not enter the stress tensor through pair separations at all. That is a real limit, stated as one, and what it does is move the question rather than leave it vague: PST has mass elsewhere, in the elementary mass $m_{0}=4\pi^{3}r_{0}^{2}d_{0}/(3c^{2})$ of §9.5 and in the Higgs sector, so what survives is how the matter action rather than the pair geometry carries it into $T_{\mu\nu}$.

7.5.  The field equations as a conservation statement

Einstein’s field equations then read:

$$G_{\mu\nu}\;=\;8\pi G\,T_{\mu\nu},\qquad T_{\mu\nu}\;=\;\Lambda\,g_{\mu\nu}\;+\;T^{\mathrm{(fields)}}_{\mu\nu}$$ (48)

The first term is the vacuum: conservation forces the scalar projection $\Pi(T(C))$ of (39) to be constant, so it is exactly $\Lambda g_{\mu\nu}$ and can be nothing else (Computation 155). The second is the stress-energy of the projected matter fields, $T^{\mathrm{(fields)}}_{\mu\nu}=-2(-g)^{-1/2}\,\delta S_{\mathrm{matter}}/\delta g^{\mu\nu}$, the same object (51) obtains by varying (50), whose content is the Standard Model that Chapters 12 and 13 derive.

Where the pair-tension projection sits in that, and where it does not. $\Pi_{\mu\nu}$ of (40) is not a third term and is not the whole of $T^{\mathrm{(fields)}}_{\mu\nu}$ either. It is that tensor’s conformal part: Computation 158 and Computation 159 show it is traceless for homogeneous matter and reproduces the radiation sector exactly, $w=1/3$, while carrying no rest mass at all. Rest mass reaches $T_{\mu\nu}$ through the matter action, by way of the Higgs and Yukawa structure of Chapter 14, and not through pair separations. Writing $T_{\mu\nu}=\Pi(T(C))\,g_{\mu\nu}$, or $\Lambda g_{\mu\nu}+\Pi_{\mu\nu}$, would say that the universe contains a cosmological constant and radiation and nothing besides: no matter era, no structure formation. It contains those because $T^{\mathrm{(fields)}}_{\mu\nu}$ does, for the ordinary reason that a massive field in its non-relativistic regime is pressureless.

The conformal identification, and it now has a criterion rather than a caveat. PST’s ambition here is that geometry and matter are coprojections of one tension, and the operator $\Pi$ is what carries that claim. Computation 159 showed $\Pi$ delivers the conformal part and cannot deliver the rest. Computation 164 goes further and says exactly which conformal tensors it delivers, in both directions, which is a better position than the caveat it replaces.

The criterion. Every term of (40) is a rank-one dyad $\Delta_{\mu}\Delta_{\nu}$ carrying a weight $W(s)=-|s|$ that never changes sign, times a bond kernel that is a density. A conic sum of semidefinite matrices with same-signed coefficients is semidefinite, so $\Pi_{\mu\nu}$ is always semidefinite, and conversely every semidefinite target is a sum of dyads. The reachable set is exactly the semidefinite metric-traceless cone, sharp in both directions, and confirmed over six ensembles, an isotropic Gaussian cloud and five deliberately awkward ones: flattened to a plane, stretched along time, split into two clumps, laid on a light cone, collapsed onto a line. Not one negative eigenvalue. That is the pair sum itself; what the identification concerns is its conformal part, which reaches further, and the rest of this section says how far.

What that class contains is everything this book rests on here. Isotropic radiation, $\mathrm{diag}(\rho,\rho/3,\rho/3,\rho/3)$, is positive definite. A null plane wave, $\Phi\,k_{\mu}k_{\nu}$, is rank one and semidefinite. So Computation 158’s $w=1/3$, Computation 159’s kinetic law, Computation 160’s polarisation count and Stefan–Boltzmann constant and Computation 161’s Planck spectrum all live inside the reachable class, and none of them is touched by what follows.

And one exclusion stops being a design choice. $\Lambda g_{\mu\nu}$ is indefinite, so the pair sum could not have produced it whatever one wanted. The separation of the vacuum term into its own sector, which §16.5 argues on physical grounds, is forced by the structure as well.

What lies outside, stated plainly. A static magnetic field has $T_{\mu\nu}=\mathrm{diag}(u,u,u,-u)$ with $u=B^{2}/2$: metric-traceless, and so conformal by this book’s own classification, but indefinite, the negative eigenvalue being the tension along the field lines. An electric field is the same. No pair sum equals it, and that was for a time the first structural item. The reason is physical rather than technical: a bond under a single-signed weight either pushes or pulls throughout, and a magnetic field does both at once, tension along the lines and pressure across them.

The one freedom left after Computation 163, a constant metric-traceless shift, does restore semidefiniteness, so what fails is the identification as stated rather than a gauged version of it. The price is that the constant is universal, must dominate the largest tension anywhere, and is not fixed by anything in the construction, since $\rho$ is a quotient map onto $M$ and leaves no empty region to pin it against.

And the obstruction does not reach the identification (Computation 199). The identification is a statement about conformal parts, and this book defines the conformal part of a tensor as the tensor minus a quarter of its metric trace times the metric. Applied to the pair sum term by term, each dyad becomes $\Delta_{\mu}\Delta_{\nu}-\tfrac{1}{4}\,s\,g_{\mu\nu}$, traceless identically and indefinite: a spacelike bond carries pressure along itself and tension across itself. Two families of bonds in the plane transverse to a direction sum to exactly $\mathrm{diag}(u,u,u,-u)$ with the field along that direction, and the electric tensor is the same construction in the other transverse plane; both are reached to machine precision from explicit clouds. The reachable set is sharp in both directions again, and larger: the closed cone the traceless dyads span is exactly the metric-traceless tensors obeying the null energy condition, $T_{\mu\nu}k^{\mu}k^{\nu}\geq 0$ for every null $k$. Inclusion is one line, $(\Delta\cdot k)^{2}\geq 0$ with the metric term killed by $k$’s nullity; equality is a duality argument, the dual cone being the positive semidefinite metric-traceless matrices, whose extreme rays are rank one by Pataki’s bound on spectrahedra and therefore null dyads, the rank-two case checked by hand on the two null directions of its plane. Every classical field obeys the condition, so every conformal part a field of this book produces is reachable, and what is excluded is exactly a null-energy violation: a traceless tensor with negative energy density along some null direction, which nothing here contains. On every balanced cloud the conformal part is the pair sum itself, the trace being Computation 158’s balance scalar and vanishing there, so Computation 158 to Computation 161 are untouched; out of balance the sum carries a trace the identification does not use, and Computation 159 had already met it: the worldline sector of the pair sum is exactly kinetic, $w=\beta^{2}/3$, dust at rest, and it carries a few per cent of the density for homogeneous matter, which is the share by which the projection there is conformal rather than exactly traceless. The forced weight is load-bearing in a new way: a weight changing sign across the light cone would flip the spacelike dyads and the cone would reach null-energy violations, so $W(s)=-|s|$ is what confines the pair sum to physical matter. The universal constant shift above is no longer needed, and the magnetic tensor sits on the boundary of the cone, the condition saturating along the field lines, as a static field should.

And a second thing is free, which belongs beside the first. The weight is forced and the bond kernel normalised, so the only remaining freedom in $\Pi_{\mu\nu}$’s magnitude is how many pairs there are, and the theory does not supply that: $d_{0}$ is a coherence length, which §15.3.1 says is not the microscopic grain, and $\ell_{*}$ enters as a momentum cutoff on field modes. Reading the spacing off one rather than the other moves the implied density by $(d_{0}/\ell_{*})^{6}\approx 10^{61}$ at the Casimir bound (Computation 167; $9\times 10^{6}$ at the derived $d_{0}$, Computation 201). Nothing claimed here reads that scale, which is why nothing claimed here is affected: every result the pair sum has produced is a ratio, a shape or a count, and Computation 167 verifies the invariance rather than assuming it. One consequence deserves stating plainly, because a reader will reach for it: there is no vacuum-energy catastrophe in this sector. That problem needs a computed scale to disagree with measurement, and this construction does not produce one.

The density is not missing, though, and saying so changes the shape of the item (Computation 179). Computation 167 searched among the theory’s lengths and found neither to be a spacing, which is right as far as it goes; A1 names one outright. §18.1 introduces $a_{n}$ as “the substrate site spacing, distinct from the fixed coherence length $d_{0}$”, with $|D_{n}|a_{n}^{4}\to V_{M}$, $a_{n}\to 0$ and local node density $\lambda_{n}=1/a_{n}^{4}\to\infty$, and it scales the same substrate: (40) sums over pairs of $D$ through $\rho$. So the density is supplied and it is infinite. With the pair cutoff the book itself uses, $|s|\lesssim d_{0}^{2}$ at the fixed coherence length (Computation 159), a cutoff that does not scale against a density that does gives $\Pi_{\mu\nu}\sim|D|^{2}\sim a_{n}^{-8}$, measured, and Computation 182 confirms that the exponent survives the bond kernel’s localisation rather than being an artefact of summing globally.

Which $D$ that is, is not a choice (Computation 182). It reads as one, between A1’s sequence and one physical substrate of finite spacing, and the second is not available. $D$ is finite at every stage, by four independent commitments of this book: A1’s own $|D_{n}|\to\infty$, the Bernoulli product $\bigotimes_{a\in D}$, the exact enumeration over $2^{|D|}$ configurations of §10.3.2.5, and the balance probability $\binom{|D|}{|D|/2}2^{-|D|}$ of §3.5. A finite set has no uncountable quotient, so the quotient $M=D/{\sim_{\rho}}$ of §4.10 is finite at every stage and is a smooth four-manifold at none of them, while (46) takes a covariant derivative of $\Pi_{\mu\nu}$ on a smooth $M$ and (41) decomposes it against a metric. So $D$ in (40) is A1’s sequence, and $M=\mathbb{R}\times S^{3}$ is the limit of the finite quotients rather than the quotient of any one substrate. Computation 167’s arithmetic is reproduced exactly and is not in question; what falls is its reading of the symbol, and Computation 179 recorded that as an unstated premise where the sharper statement is that it was an unavailable one.

And the forced normalisation does not buy a finite density, which is the correction this makes to Computation 179. Dividing by the pair count is forced and it converges. But (40) is dimensionless: $W(s)$ carries $L^{2}$, the unnormalised dyad another $L^{2}$, and a bond kernel normalised to unit integral carries $L^{-4}$, so the bare sum is a pure count, which is why it goes as $|D|^{2}$. A stress tensor is an energy density, and the only scale the discrete data carries is $a_{n}$, so restoring the dimensions costs exactly one factor $a_{n}^{-4}$, which is $\lambda_{n}$ itself. Measured, the restored object diverges as one power of the node density, fitted at $1.011$ against controls at $0.049$ and $2.049$ (Computation 182).

So M7’s missing quantity is named, and there is exactly one of it: the physical spacing $a_{*}$ at which A1’s idealised limit is truncated, with $\Pi_{\mu\nu}\sim 1/a_{*}^{4}$. That is the same object a lattice spacing is, and it puts this construction in the ordinary position of a lattice one: the continuum limit does the geometry, the finite spacing sets the magnitude, and no result that is a ratio, a shape or a count feels it. Computation 167’s finding that neither $d_{0}$ nor $\ell_{*}$ is such a spacing stands, which is why the item does not close; what changes is that the gap is one truncation scale of a familiar kind rather than a density the theory was said not to supply. The vacuum-energy conclusion above is unaffected, since it needs only that no absolute scale is computed, and $a_{*}$ is named rather than computed.

And $a_{*}$ is not a constant of the theory (Computation 196). On the Friedmann geometry of Chapter 16, (46) with the trace zero makes a homogeneous traceless $\Pi_{\mu\nu}$ fall as $a(t)^{-4}$, for every spatial curvature; a fixed physical $a_{*}$ is the case $\rho=\mathrm{const}$, whose divergence is $4H\rho$, so the truncation spacing is comoving. Under the identification this section makes, the traceless homogeneous part of $T^{\mathrm{(fields)}}_{\mu\nu}$ is the radiation and nothing else ($\rho-3p=0$ iff $w=1/3$), so $\Pi^{0}{}_{0}=\rho_{r}$ and $a_{*}=(\hbar c/\rho_{r})^{1/4}=0.82$ mm today, which is $0.97$ times $\hbar c/kT_{\mathrm{CMB}}$: the thermal wavelength of the radiation bath, scaling as $a(t)$. It passes through the Casimir bound on $d_{0}$ at $kT=83$ eV, through the derived $d_{0}$ at $kT\approx 40$ GeV (Computation 201) and through $0.4\,\ell_{*}$ at $kT=M_{*}$, which is Computation 167’s negative result as a function of time. So the magnitude of $\Pi_{\mu\nu}$ is supplied by the state, as the magnitude of $T_{\mu\nu}$ is in general relativity; with Computation 199 removing the cone restriction the identification has no gap left and is carried in Appendix C.6; the vacuum-energy conclusion is untouched because $\rho_{r}$ is read, not computed.

This is the same unstated choice Computation 177 found in P3, in a second construction. There (55) was a bare sum over $D$ while (57) divides by $|D|$, so a constant coefficient diverged under A1 until Computation 177 showed $c$ must carry a $1/|D|$; here (40) is a bare sum over pairs of $D$ carrying no normalisation at all. Every bare sum over $D$ is sensitive to which reading is meant, and that is worth checking in any further construction of this kind.

A second doubling sits underneath it and is worth stating separately, because it is about a definition rather than a sum. This book gives the realisation map twice and the two are not of the same type. §4.10 makes $\rho$ a surjective quotient whose target is derived, “$M$ is the quotient set $D/{\sim_{\rho}}$”, with a spacetime point defined to be a fibre, and it spends the surrounding paragraphs establishing that $M$ is not a container given in advance. §18.1 then writes $\rho_{n}:D_{n}\to M=\mathbb{R}\times S^{3}$ with image “asymptotically equidistributed”, which for finite $D_{n}$ cannot be surjective and takes $M$ as given; and the assumption list of §7.8 says as much outright, that $\rho$ “presumes the target manifold $M$ is smooth and finite-dimensional”. The reconciliation is the one above: the quotient is the stage-wise statement and the smooth $M$ is the limit. The fibre structure, which an earlier printing named here as unsaid, is settled on the theory’s own measure (Computation 212): a fibre is a clique of pairwise-null properties, a set of directions in $V_{7}$ with one pairwise inner product, so it holds at most seven properties (the Gram rank), and under the $G_{2}$-invariant law on directions an exact null coincidence has probability zero, so fibres are singletons almost surely. $|D_{n}|a_{n}^{4}\to V_{M}$ counts properties against volume, and A1 reads $|D_{n}|$ as the point count; the exponent above is insensitive to the fibre size in any case, and §4.10’s identification of a point with a bundle of properties holds with a bundle of size one, at most seven (Computation 182).

So the honest statement is this, and it is a stronger one than before. $\Pi_{\mu\nu}$ is the conformal part of $T^{\mathrm{(fields)}}_{\mu\nu}$: on the semidefinite sector by the characterisation above, which contains every result claimed here, and beyond it by Computation 199, which removed the cone restriction, so the identification has no gap left and no universal constant to supply. A framework that can say exactly which part of $T_{\mu\nu}$ it derives, and from what, is in a better position than one that takes the whole of $T_{\mu\nu}$ as input and never reaches the question. General relativity is the second kind.

7.5.1.  Why both routes are kept

The question this invites is whether the projection is then redundant. If the matter action supplies $T^{\mathrm{(fields)}}_{\mu\nu}$, and supplies exactly the part $\Pi$ cannot, what work is left for $\Pi$ to do? Three things, and the matter action does none of them.

It supplies $\Lambda$, and the matter action cannot. The vacuum energy of $S_{\mathrm{matter}}$ is the Standard Model’s, with the hierarchy problem §16.5 says plainly PST inherits. What the projection adds is a term of a kind no field on the manifold can produce: $\rho_{\mathrm{inst}}=j_{\tau}\,\Delta\mathcal{F}/c^{2}$ of (221) does not dilute, because the configuration space it draws on is precausal and has no spatial extent to stretch. That fixes the sign of $\Lambda$, positive and de Sitter, from the direction of every threshold crossing, and fixes $w=-1$ as a theorem rather than a fit (Computation 137, Computation 155). The magnitude stays $T(C)$-contingent and is not claimed. Without the projection, $\Lambda$ is once again a number to be fitted, which is the position PST set out to improve on.

It makes the radiation sector a derivation rather than an input. In the matter action, $w=1/3$ holds because a massless field with a conformal action was written down; the answer was already in the premise. In the projection it holds because the derived weight $W(s)=-|s|$ of (43) vanishes identically on null separations, and the separation that is null is the one $\tau_{\mathrm{null}}$ defines in (30). §8.3.4 carries that further: the same threshold fixes the two polarisations, the Stefan–Boltzmann law and its coefficient. Those are facts about light obtained from the object that fixes the signature, and $S_{\mathrm{matter}}$ on its own reaches none of them.

It is where the unification claim has content, and it is checkable. A formulation with the matter action alone is General Relativity with a prior account of the manifold: geometry from the substrate, matter handed in from outside. That is less than PST claims. And because $\Pi_{\mu\nu}$ is identified with the conformal part of $T^{\mathrm{(fields)}}_{\mu\nu}$ rather than with all of it, the identification is a statement about a sector that is independently known, so it can fail: the traceless part of the matter stress tensor has to equal the pair-tension sum (40). A formulation without the projection has nothing of that kind to fail.

So the two are not rival suppliers of the same tensor, one of which has been made redundant. The projection’s job has changed: from carrying the whole right-hand side, which Computation 159 rules out, to carrying the vacuum term, carrying the conformal sector, and constraining the rest.

The standard reading of this equation is dynamical: matter curves spacetime, and the degree of curvature is set by the amount of matter. The PST reading is different and more fundamental: the field equations are a conservation statement expressing that the total asymmetric tension of the precausal configuration is faithfully preserved across the sublimation threshold, distributed simultaneously as curvature on the left-hand side and as stress-energy on the right. The equations do not govern a process; they describe what the functor $\Phi$ conserves. Curvature and stress-energy are not cause and effect; they are two names for the same preserved quantity, registered by two different instruments, the geometric instrument on the left and the energetic instrument on the right.

This reinterpretation connects with, and goes beyond, the thermodynamic approach of Jacobson [19] and the entropic gravity programme of Verlinde [18]. Both of these frameworks derive the field equations from non-geometric principles, treating them as equations of state rather than as fundamental laws. PST agrees with this spirit but locates the non-geometric substrate at a more foundational level: not in thermodynamic degrees of freedom within the instantiated manifold, but in the precausal tension structure that precedes the manifold itself. Where Jacobson derives Einstein’s equations from entropy on holographic screens, PST derives them from the structure of $\Phi$ and $\Pi$. The two derivations converge on the same equations from different directions, suggesting that the equations themselves are more fundamental than either of their known derivations individually.

7.6.  The F–S dichotomy: free energy versus action

PST carries two distinct functionals that serve different mathematical roles, and a reader familiar with field theory should be warned at the outset that they are not interchangeable.

$\mathcal{F}[\psi,C]$ is the modal free energy: a Landau–Ginzburg potential defined over precausal configuration space. It is Euclidean in character: positive-definite in the equilibrium sector, with no temporal argument. Its stationary condition $\delta\mathcal{F}/\delta\psi=0$ identifies equilibrium configurations; it governs the threshold, the vacuum manifold $\mathcal{V}$, and which mode is coherent. No time appears; no Noether theorem applies. $\mathcal{F}$ is not an action and must not be treated as one.

$S[g,\phi]$ is the instantiated action: a standard Lorentzian functional defined over spacetime histories in the instantiated domain $\mathcal{G}$. Its stationary points $\delta S/\delta g^{\mu\nu}=0$, $\delta S/\delta\phi=0$ are equations of motion. Noether’s theorem, conservation laws, and all dynamical structure belong here. $S$ is a projected object: it does not exist in the precausal substrate; it is what $\mathcal{F}$ becomes under the action of $\Pi$, with the Bernoulli measure $d\mu$ mapping to the covariant measure $\sqrt{-g}\,d^{4}x$ and the gradient term $c|\nabla_{\mathcal{C}}\psi|^{2}$ projecting to the kinetic sector of $\mathcal{L}$.

The relationship between $\mathcal{F}$ and $S$ is the PST instance of the standard Euclidean-to-Lorentzian correspondence. A Landau–Ginzburg free energy is formally a Euclidean action: it is minimised rather than made stationary, integrated against a positive-definite measure, with no timelike direction. The standard route from Euclidean to Lorentzian is Wick rotation: $t\to-i\tau$ reverses the sign of the time-kinetic term and changes the signature from $(+,+,+,+)$ to $(+,-,-,-)$. In PST, sublimation is the Wick rotation. The moment the $(+,-,-,-)$ Lorentzian signature emerges from the functor $\Phi$ is precisely the moment $\mathcal{F}$ ceases to be the relevant functional and $S$ takes over. This is not an embarrassing gap in the formalism. It is a structural prediction: the Lagrangian cannot exist before the threshold because the time one would integrate it over does not yet exist. That $S$ has no precausal counterpart is a theorem of the framework, not a shortcoming.

F–S Dichotomy. $\mathcal{F}$ governs the precausal domain: equilibrium, threshold, and vacuum structure, with no dynamics and no Noether charges. $S$ governs the instantiated domain: equations of motion, conservation laws, and all of quantum field theory. The passage from $\mathcal{F}$ to $S$ is sublimation; it is not reversible; and nothing on the $S$-side can be traced back to the precausal level without crossing the threshold. Every subsequent use of Noether’s theorem, every conserved charge, and every field-theoretic result in this book belongs exclusively to $S$.

7.7.  The projected action and the Einstein–Hilbert term

Newton’s constant from the spectral action. Newton’s constant is the gravitational coefficient of the spectral action $\mathrm{Tr}\,f(D/\Lambda)$ [49]. Its heat-kernel asymptotics contribute a cosmological term at order $\Lambda^{4}$, the Einstein–Hilbert term $\tfrac{1}{16\pi G}\int R\sqrt{-g}$ at order $\Lambda^{2}$, and the Yang–Mills [155] and Higgs terms at order $\Lambda^{0}$. The Seeley–DeWitt heat-kernel computation [57] carried out in this section and the next is exactly that $\Lambda^{2}$ coefficient: $1/G\sim f_{2}\,\Lambda^{2}\,\mathcal{N}$, with $\mathcal{N}$ the trace over the finite internal factor and the cutoff $\Lambda\sim 1/d_{0}$ set by the Landau–Ginzburg coherence scale. The scales $r_{0}$ and $d_{0}$ are the Landau–Ginzburg vacuum scales of the order parameter, and $G$ emerges as a derived consistency relation among the observables $(m_{h},M_{P},v)$ rather than a first-principles output.

The field equations (48) were presented above as a conservation statement, but their specific tensor structure (why the left-hand side is $G_{\mu\nu}$ and not $R_{\mu\nu}$, or $R\,g_{\mu\nu}$, or some other curvature combination) requires a derivation that has so far been deferred. That derivation begins with the projected action.

The potential functional $\mathcal{F}(\psi,C)$ of equation (77) (stated in full later) contains three terms. The first two, $-\varepsilon\,|\psi|^{2}$ and $b|\psi|^{4}$, determine the equilibrium structure of the order parameter. The third, $c\,|\nabla_{\mathcal{C}}\psi|^{2}$, is the gradient term: it penalises variation of the order parameter across configuration space and controls the energetic cost of excitations above the vacuum. Under the projection $\Phi$, this term acquires a precise geometric interpretation.

When the order parameter has settled to its vacuum value $\psi=r_{0}\,e^{i\theta}$ on $\mathcal{V}$, the configuration-space gradient $\nabla_{\mathcal{C}}\psi$ decomposes into two pieces: the Goldstone mode $r_{0}\,\nabla_{\mathcal{C}}\theta$ (responsible for matter and angular momentum, discussed below) and a metric fluctuation component, which encodes how the instantiated geometry responds to variations of the realisation map $\rho$. Integrating the metric fluctuation component over $\mathcal{C}$ with the measure $\mu$ and projecting onto $(M,g)$ via $\Pi$, the leading term in a derivative expansion of the projected gradient is:

$$\Pi\!\left(c\,|\nabla_{\mathcal{C}}\psi|^{2}\right)\big{|}_{\psi\,=\,r_{0}}\;\longrightarrow\;\frac{r_{0}^{2}\,c}{16\pi G}\int_{M}R\,\sqrt{-g}\;\mathrm{d}^{4}x\;+\;\mathcal{O}\!\left(R_{\mu\nu}R^{\mu\nu},\,R^{2}\right)$$ (49)

where $R$ is the Ricci scalar of $(M,g)$. The coefficient $1/(16\pi G)$ sets the gravitational coupling scale and identifies $G$ as derived from $r_{0}$, $c$, and the geometry of the projection: consistent with equation (72) (derived later, §8.9). The higher-order terms in $R_{\mu\nu}R^{\mu\nu}$ and $R^{2}$ are suppressed by $(k\,d_{0})^{2}$ and are negligible at all scales $\gg d_{0}$. The effective projected action at low energies is therefore:

$$S[g]\;=\;\frac{1}{16\pi G}\int_{M}R\,\sqrt{-g}\;\mathrm{d}^{4}x\;+\;S_{\mathrm{matter}}[g,\,\text{fields}]$$ (50)

This is the Einstein–Hilbert action. The step from the gradient term in $\mathcal{F}$ to equation (50) is structural. The functional form of the projection (that the leading curvature term is $\int R\sqrt{-g}\,\mathrm{d}^{4}x$ and not a higher-curvature combination) is derived in §7.8 via the Seeley–DeWitt heat-kernel expansion. That resolves the projection-coefficient question: the structure of the R term is established here, and its numerical coefficient $\mathcal{N}$ is identified as the finite-internal-factor trace in §7.1. The downstream Noether–Lovelock argument [22] is standard and unambiguous given equation (50). Varying equation (50) with respect to $g^{\mu\nu}$ yields, by a standard calculation:

$$\frac{\delta S}{\delta g^{\mu\nu}}\;=\;0\qquad\Longrightarrow\qquad G_{\mu\nu}\;=\;8\pi G\;T_{\mu\nu}$$ (51)

where $T_{\mu\nu}={-2}/{\sqrt{-g}}\cdot\delta S_{\mathrm{matter}}/\delta g^{\mu\nu}$ is the stress-energy tensor of the projected matter fields. The Einstein tensor $G_{\mu\nu}=R_{\mu\nu}-\tfrac{1}{2}R\,g_{\mu\nu}$ appears specifically because it is what the variation of $\int R\sqrt{-g}$ with respect to $g^{\mu\nu}$ produces, up to boundary terms. Equation (48) is thereby derived, not asserted.

7.8.  Seeley–DeWitt argument for the projected action

The previous section identified the projection coefficient $\mathcal{N}$ in equation (49) as the trace over the finite internal factor. This section shows that the functional form of the projection (the leading curvature term $\int R\sqrt{-g}\,\mathrm{d}^{4}x$ announced at equation (50)) is forced by a universal result from differential geometry: the Seeley–DeWitt heat-kernel expansion [57, 59]. The numerical coefficient (which fixes $G$ in terms of $r_{0}$, $c$, and $d_{0}$) is delivered by the spectral-action moments of Computation 5, $1/(16\pi G)=(f_{2}\Lambda^{2}/\pi^{2})\cdot\kappa_{\mathrm{grav}}\cdot N_{\mathrm{dof}}$, with the matched cutoff $\Lambda$ and the fermionic multiplicity $N_{\mathrm{dof}}$ as the calibration inputs. Pinning $G$ to its observed value via this relation fixes one cutoff combination at the Planck scale (consistent with the $M_{*}=4\pi m_{h}\sqrt{2/3}$ matched scaling at the electroweak sector); the numerical coefficient is therefore a calibration of the cutoff identification rather than an open derivation, on the same footing as the coherence length $d_{0}$ (structurally a condensate quantity, $2.2$ am under the canonical normalisation, Computation 201, and bounded by $2.3$ nm from the data via (217)).

Setup. The projection operator $\Pi$ is characterised by a kernel $K\colon\mathcal{C}\times M\to\mathbb{R}$ satisfying

$$\Pi(A)[x]\;=\;\int_{\mathcal{C}}A[\psi]\,K(\mathcal{C},x)\,\mathrm{d}\mu(\mathcal{C})$$ (52)

for any functional $A$ on configuration space. The kernel inherits the permutation invariance of the Bernoulli measure $\mu$; its spatial covariance is controlled by the substrate coherence length $d_{0}$, so that $K(\mathcal{C},x)$ couples configurations within a neighbourhood of radius ${\sim}\,d_{0}$ of the pre-image $\rho^{-1}(x)$.

Vacuum decomposition. At the vacuum $\psi=r_{0}\,e^{i\theta}$, the configuration-space gradient decomposes as

$$|\nabla_{\mathcal{C}}\psi|^{2}\;=\;r_{0}^{2}\,|\nabla_{\mathcal{C}}\theta|^{2}\;+\;|\nabla_{\mathcal{C}}\sigma|^{2}\;+\;\mathcal{O}(\sigma^{2})$$ (53)

where $\sigma=|\psi|-r_{0}$ is the amplitude (metric-fluctuation) mode and $\theta$ is the Goldstone (phase) mode. The Goldstone term $r_{0}^{2}|\nabla_{\mathcal{C}}\theta|^{2}$ projects to the gauge and matter sector (Chapters 13 and 14). The amplitude term $|\nabla_{\mathcal{C}}\sigma|^{2}$ encodes how the instantiated geometry responds to deformations of the realisation map $\rho$: a deformation $\delta\rho$ induces a metric variation $\delta g_{\mu\nu}=\mathcal{L}_{\delta\rho}g_{\mu\nu}$ (the Lie derivative), and $\nabla_{\mathcal{C}}\sigma$ in the direction of $\delta\rho$ is the corresponding functional derivative of the amplitude. It is this term that, when projected via $\Pi$, produces the gravitational action.

The heat-kernel form of the projected integral. The projected amplitude gradient can be written as a smeared trace over $M$:

$$\Pi\!\left(c\,|\nabla_{\mathcal{C}}\sigma|^{2}\right)\bigl{|}_{\sigma=0}\;=\;c\,r_{0}^{2}\int_{M}\widetilde{K}(x,x;\,d_{0}^{2})\,\sqrt{-g}\;\mathrm{d}^{4}x$$ (54)

where $\widetilde{K}(x,y;\,t)$ is the coincidence limit of the projection kernel restricted to metric fluctuations, with $t=d_{0}^{2}$ playing the role of the “heat-kernel time.” The Seeley–DeWitt theorem requires $\widetilde{K}$ to be, to leading order, the heat kernel of a Laplace-type operator $P=-(g^{\mu\nu}\partial_{\mu}\partial_{\nu}+\cdots)$ whose principal symbol is the spacetime metric $g$. This is not a consequence of smoothness, positivity, symmetry, and localisation: a normalised geodesic Gaussian satisfies all four properties yet has diagonal curvature coefficient $\tfrac{1}{3}R$ rather than $\tfrac{1}{6}R$, as a direct calculation on $S^{2}$ confirms. The $\tfrac{1}{6}$ is a property of the heat kernel of the Laplacian specifically, not of the broad class of localised kernels.

The Laplace-type condition is derived from P1–P3 via Mosco convergence (§7.8), and stated separately in §1.7 for reference. It runs as follows. $\widetilde{K}$ is defined as the heat kernel of the projected Boolean Laplacian $\Pi(L)$, with the metric $g$ defined as the principal symbol of $\Pi(L)$. This breaks the potential circularity: $g$ is not presupposed when defining $\widetilde{K}$; it is read off from the operator. With this identification the Seeley–DeWitt expansion applies and the universal coefficient $a_{1}=\tfrac{1}{6}$ forces the leading curvature term to be $\int R\sqrt{-g}\,\mathrm{d}^{4}x$.

Taken as an independent postulate, the Laplace-type projection condition would be non-trivial: stipulating that $\Pi(L)$ is Laplace-type is close to postulating the kinematic skeleton of a metric field theory directly. The Mosco-convergence argument of the following paragraph derives this condition from P1–P3 as a theorem. The honest reading of PST’s gravitational sector is therefore that General Relativity is recovered from three foundational postulates, with the operator-type condition following as a structural consequence of Bernoulli isotropy.

Deriving the Laplace-type condition from the Bernoulli measure (Mosco convergence). The Laplace-type projection condition is derived by an argument using Mosco convergence of Dirichlet forms [98]. Define the Boolean Dirichlet form on $L^{2}(\mathcal{P}(D),\mu)$ by

$$E^{B}(u,v)\;=\;\sum_{a\in D}\int_{\mathcal{P}(D)}(\partial_{a}^{B}u)(C)\,(\partial_{a}^{B}v)(C)\,\mathrm{d}\mu(C),$$ (55)

where $\partial_{a}^{B}$ is the Boolean difference operator of equation (20) and the generator of $E^{B}$ is the Boolean Laplacian $L^{B}$. For smooth functions $f,h\in C^{\infty}(M)$ pulled back via $\Pi^{\dagger}$, the isotropy of the Bernoulli measure $\mu$ forces the covariance matrix of the displacement $\rho(a)^{\mu}-x^{\mu}$ to be proportional to $g^{\mu\nu}(x)$:

$$\frac{1}{|D|}\sum_{a\in D}\bigl{(}\rho(a)^{\mu}-x^{\mu}\bigr{)}\bigl{(}\rho(a)^{\nu}-x^{\nu}\bigr{)}\;\longrightarrow\;\tfrac{d_{0}^{2}}{4}\,g^{\mu\nu}(x)$$ (56)

(no preferred direction exists in the substrate, so the only $O(4)$-covariant rank-2 tensor built from $g$ is $g^{\mu\nu}$ itself). The isotropy here is Euclidean, and it fixes only the positive-definite content: the scale $\tfrac{d_{0}^{2}}{4}$ and the absence of a preferred spatial direction. A covariance is positive-semidefinite and cannot itself carry indefinite signature. The Lorentzian signature of $g$ is therefore supplied separately by the $\tau_{\mathrm{null}}$ sign cut ((30), Computations 2 and 124), and (56) is written in that post-signature metric. It follows that

$$\frac{1}{|D|}\,E^{B}(\Pi^{\dagger}(\Pi f),\,\Pi^{\dagger}(\Pi h)) \xrightarrow[|D|\to\infty]{\;\mathrm{Mosco}\;}\;\frac{d_{0}^{2}}{4\,V_{M}}\int_{M}g^{\mu\nu}\,\partial_{\mu}f\,\partial_{\nu}h\;\sqrt{-g}\;\mathrm{d}^{4}x$$ (57)

$$=:\;\frac{d_{0}^{2}}{4\,V_{M}}\,E^{\mathrm{Riem}}(f,h),$$

where $E^{\mathrm{Riem}}$ is the Riemannian Dirichlet form whose generator is the Laplace–Beltrami operator $\Delta_{g}$. The $1/V_{M}$ is the same normalisation A2 carries, and it is not optional: the left side is an expectation against the probability measure $\mu$ while $E^{\mathrm{Riem}}$ as written is an unnormalised integral, so the two cannot be equated without it, and Computation 177 measures the factor on an explicit lattice and recovers it to $0.3\%$ across a factor $81$ in volume. As with A1 and A2 the statement is read locally, $V_{M}$ being infinite on $M=\mathbb{R}\times S^{3}$, so $V_{M}$ is $\mathrm{vol}(K)$ on the compact $K$ carrying the test function; equivalently, and more cleanly, the limit is $\tfrac{d_{0}^{2}}{4}\langle g^{\mu\nu}\partial_{\mu}f\,\partial_{\nu}h\rangle$ with $\langle\cdot\rangle$ the normalised average. By Mosco’s theorem [98], this $\Gamma$-convergence of quadratic forms implies strong resolvent convergence of their generators:

$$\bigl{(}|D|^{-1}L^{B}+\lambda\bigr{)}^{-1}\Pi^{\dagger}\;\xrightarrow{\;|D|\to\infty\;}\Pi\bigl{(}\Delta_{g}+\lambda\bigr{)}^{-1}\qquad\forall\,\lambda>0.$$ (58)

The generator of the limiting form is $\Delta_{g}$, a Laplace-type operator with principal symbol $g^{\mu\nu}$, so $\Pi(L^{B})$ is Laplace-type in the scaling limit: the projection condition follows from P1 (the distinction primitive, with the Bernoulli measure derived in Chapter 2, §2.5), P2 (asymmetric tension) and P3 (the Landau–Ginzburg modal potential), within the scaling argument.

What is assumed at the Mosco step. The argument above derives the Laplace-type character of $\Pi(L^{B})$ in the scaling limit; it does not derive the dimensionality $n=4$ or the Lorentzian signature $(+,-,-,-)$ from $S_{D}$-symmetry of the Boolean substrate alone. Three structural elements enter at the realisation step beyond P1–P3:

1. The realisation map $\rho:D\to M$ presumes the target manifold $M$ is smooth and finite-dimensional.

2. The dimensionality $n=4$ enters via the index structure $\rho(a)^{\mu}-x^{\mu}$ ($\mu=0,1,2,3$) and the notation $O(4)$ in the isotropy step. $S_{D}$-symmetry on the discrete substrate by itself does not specify the target dimension, and the natural large-$|D|$ limit of the Hamming cube $\{0,1\}^{|D|}$ is an infinite-dimensional Gaussian object unless an additional structural element reduces it.

3. The Lorentzian signature is the sign cut at the null threshold $\tau_{\mathrm{null}}$ in equation (30).

Under the native reading these three are not free posits but follow from the verified Connes-spectral-triple structure of the substrate. Computation 3 shows that on the parity-selected sub-limit (odd $|D|$) the substrate’s complementation-and-parity real structure makes it a real Connes spectral triple of real K-theory (KO) dimension six with the Lorentzian-compatible sign triple $(+,+,-)$. Computation 4 shows that the Mosco-limit spacetime $M=\mathbb{R}\times S^{3}$ has a unique spin structure (KO-dimension four). KO additivity then forces the total to $4+6=10\equiv 2\pmod{8}$, the Connes Lorentzian Standard-Model value. This is not an assumption but the verified arithmetic of the product spectral triple; $n=4$ is forced because any other value would either contradict the substrate’s parity-verified KO-6 or fail KO additivity to ten. Lorentzian signature follows by Computation 2 hyperbolic well-posedness of the Goldstone wave equation (§4.12); the null threshold $\tau_{\mathrm{null}}$ is the sign cut that hyperbolicity requires, not an independent input. PST therefore derives the dimension and signature of $M$ from the three postulates plus the verified spectral-triple structure of the substrate and the Mosco-limit spacetime. Deriving these directly from $S_{D}$-symmetry alone, without invoking the Connes spectral- triple machinery, would be a deeper but currently unattained result.

Lorentzian signature from the directionality of modal sublimation. A complementary physical mechanism supplies the same sign cut. The Mosco scaling limit by itself yields Euclidean signature in the $|D|\to\infty$ limit. Lorentzian signature ($+{-}{-}{-}$) arises from the directionality of modal sublimation: the tension $T(C)$ exceeds the threshold $\tau$ in a preferred configuration direction that, after instantiation, becomes the time-like direction. This breaks the $O(4)$ symmetry of the Euclidean scaling limit to the Lorentz group [181] $O(1,3)$. The Wick rotation is therefore a structural consequence of P2 (asymmetric tension) together with the threshold-crossing of P3, not an additional postulate. The Computation 2 hyperbolic well-posedness argument (§4.12) gives the same conclusion by a different route, showing it is the unique signature compatible with a well-posed Cauchy problem for the Goldstone wave equation.

The Seeley–DeWitt expansion. For any smooth kernel of this type on a four-dimensional (pseudo-)Riemannian manifold, the asymptotic expansion as $d_{0}\to 0$ gives [57, 58, 59]:

$$\widetilde{K}(x,x;\,d_{0}^{2})\;=\;\frac{\mathcal{N}}{(4\pi d_{0}^{2})^{2}}\left[1\;+\;\frac{d_{0}^{2}}{6}\,R(x)\;+\;d_{0}^{4}\,\mathcal{A}_{2}(x)\;+\;\mathcal{O}(d_{0}^{6})\right]$$ (59)

where $\mathcal{N}$ is the normalisation of the kernel (fixed by $\int_{M}\widetilde{K}\,\sqrt{-g}\,\mathrm{d}^{4}x=1$ on flat space), the coefficient $\tfrac{1}{6}$ in front of $R$ is the universal first Seeley–DeWitt coefficient $a_{1}$ for a scalar field, and $\mathcal{A}_{2}(x)=\frac{1}{180}(R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}-R_{\mu\nu}R^{\mu\nu}+6\nabla^{2}R)+\frac{1}{72}R^{2}$ is the second coefficient. The universality of $\tfrac{1}{6}$ is the key: it depends only on the spacetime dimension and the spin of the projected field, not on any property of the PST kernel $K$.

Integration and the Einstein–Hilbert structure. Substituting equation (59) into equation (54):

$$\Pi\!\left(c\,|\nabla_{\mathcal{C}}\sigma|^{2}\right)\bigl{|}_{\sigma=0} =\frac{c\,r_{0}^{2}\,\mathcal{N}}{(4\pi d_{0}^{2})^{2}}\int_{M}\!\sqrt{-g}\,\mathrm{d}^{4}x$$ (60)

$$\quad+\frac{c\,r_{0}^{2}\,\mathcal{N}}{6\,(4\pi)^{2}\,d_{0}^{2}}\int_{M}R\,\sqrt{-g}\,\mathrm{d}^{4}x\;+\;\mathcal{O}(d_{0}^{4}\,R_{\mu\nu\rho\sigma}^{2})$$

The first term is a constant (vacuum energy) that is absorbed into the cosmological constant. The second term has the precise structure of the Einstein–Hilbert action, confirming equation (49) to leading curvature order. The higher-order terms are suppressed by $(k\,d_{0})^{2}$ at momenta $k\ll 1/d_{0}$, consistent with the assertion in equation (49).

The coefficient and Newton’s constant. Comparing the coefficient of $\int R\sqrt{-g}\,\mathrm{d}^{4}x$ in equation (60) with $1/(16\pi G)$:

$$\frac{c\,r_{0}^{2}\,\mathcal{N}}{6\,(4\pi)^{2}\,d_{0}^{2}}\;=\;\frac{1}{16\pi G}\qquad\Longrightarrow\qquad G\;=\;\frac{6\,(4\pi)^{2}\,d_{0}^{2}}{16\pi\,c\,r_{0}^{2}\,\mathcal{N}}\;=\;\frac{6\pi\,d_{0}^{2}}{c\,r_{0}^{2}\,\mathcal{N}}$$ (61)

The quantity $\mathcal{N}$ is the normalisation integral of the PST kernel $K$ over the Bernoulli measure:

$$\mathcal{N}\;=\;\int_{\mathcal{C}}K(\mathcal{C},x)\,\mathrm{d}\mu(\mathcal{C})$$ (62)

evaluated at any point $x\in M$ (the result is independent of $x$ by permutation invariance of $\mu$). This defines the normalisation $\mathcal{N}$. In the spectral-triple reading, $\mathcal{N}$ is identified in §7.1 with the trace over the finite internal factor, and equation (61) then fixes $G$ as a consistency relation among $c$, $r_{0}$, and $d_{0}$ (§8.5).

Summary. The Seeley–DeWitt argument establishes three things without additional assumption. First: the functional form of the gradient projection is $\int R\sqrt{-g}\,\mathrm{d}^{4}x$ at leading curvature order, forced by the universal coefficient $a_{1}=1/6$. Second: the higher-order corrections are $R_{\mu\nu\rho\sigma}^{2}$ terms suppressed by $(k\,d_{0})^{2}$. Third: $G$ is determined by $\mathcal{N}$ via equation (61); $\mathcal{N}$ is identified in §7.1 as the trace over the finite internal factor, and $G$ is fixed as a consistency relation among $(m_{h},M_{P},v)$.

7.9.  Summary

This chapter derives energy and matter as projections of the same asymmetric tension that projects to geometry. It leans on four analytic results stated in §7.1 and proved later (three in Chapter 18, the fourth in §8.10). The stress-energy tensor is the tensor projection $\Pi_{\mu\nu}$ of the tension functional (§7.4): its scalar part is exactly the cosmological constant and its conformal part the projected matter fields, so Einstein’s equations read as a conservation statement rather than a source equation (§7.5). The modal free energy $\mathcal{F}$ and the instantiated action $S$ are distinct functionals with distinct roles; the gradient term of $\mathcal{F}$ projects to the Einstein–Hilbert term, whose functional form the Seeley–DeWitt expansion forces at leading curvature order with higher-curvature corrections suppressed by $(k\,d_{0})^{2}$ (§7.8). Newton’s constant enters as the trace over the finite internal factor and is fixed as a consistency relation among $(m_{h},M_{P},v)$, which Chapter 8 takes up.