What happens at high energy?
Inertia, amplitude, and the lattice’s built-in cutoff
The result in one paragraph
Critics of discrete spacetime reach for the same objection: a lattice has a shortest length, so it must do something unphysical at high energy — break Lorentz invariance, or hide an arbitrary cutoff, or simply stop making sense. On the A=1 lattice the answer is built into the update rule, and it is not a patch. There is only one amplitude per site, and the rate at which its phase advances, \Delta\phi=\omega+V, decides how that amplitude is split between propagating and staying in place. Staying in place is what inertia is here — it is not a second substance added to the model, it is the fraction of the amplitude the phase-rate diverts from hopping. Push the phase-rate up and the amplitude moves from the propagating channel into the inertial one; at the extreme it is fully frozen. The same phase-rate, read spatially as momentum, enters the hop through a structure factor that saturates at the zone boundary, so momentum and energy per mode are bounded. The familiar relativistic dispersion E^2=m^2+|\mathbf p|^2 is the low-energy limit of this; “high energy” does not run away, the dispersion curve bends over (the lattice itself is fixed — it is the energy–momentum relation that curves, not the substrate). None of this is bolted on — it is what the bipartite Dirac update already says.
The objection, stated fairly
The strong form of the objection is not vague. It is fermion doubling: discretise a Dirac field naively and the momentum operator, which should vanish only at \mathbf k=0, picks up extra zeros at the edges of the Brillouin zone — spurious light modes that wreck the continuum count and the chiral structure. Any lattice claiming to be the Dirac equation, rather than merely approximate it, has to say what happens at the zone boundary. So the question “what happens at high energy?” is really two questions: where does the energy you pump in actually go, and what does the dispersion do as momentum approaches the lattice scale?
What it looks like
The math
The bipartite tick rule (the operator derived in Paper I, implemented in dcl_core.core3d.HopOperator) updates the spinor as
\psi^{\text{new}}_R = \underbrace{\cos(\Delta\phi/2)\,\langle \psi_L\rangle_{\text{hop}}}_{\text{propagating}} \;+\;\underbrace{\,i\,\sin(\Delta\phi/2)\,\psi_R\,}_{\text{inertial (stay)}}, \qquad \Delta\phi=\omega+V .
The two coefficients are not free — \cos^2+\sin^2=1 — so the amplitude fractions in the two channels,
p_{\text{move}}=\cos^2(\Delta\phi/2), \qquad p_{\text{stay}}=\sin^2(\Delta\phi/2),
sum to one at every site, every tick. That sum is the \mathcal{A}=1 conservation law, seen locally: the amplitude is never created or destroyed, only repartitioned between moving and staying. At \Delta\phi=0 everything propagates — a massless photon, no inertia, moving at c every tick. At \Delta\phi=\pi everything stays — the amplitude rotates in place (i\sin, pure Zitterbewegung) and goes nowhere, the maximally inertial, heaviest mode the lattice can hold. A massive particle lives in between. Mass and inertia are the stay fraction; the phase-rate sets it.
The spatial side of the same phase-rate is momentum, \mathbf k = \nabla\phi, and it enters the hop through the structure factor
S(\mathbf k)=\tfrac12\bigl(H_{\text{RGB}}-H_{\text{CMY}}\bigr), \qquad H_{\text{RGB}}(\mathbf k)=\tfrac13\!\sum_{v\in\text{RGB}} e^{i\mathbf k\cdot v},
which is exactly HopOperator.fourier_kernel. At small \mathbf k it reduces to i\,\mathbf k\cdot\boldsymbol\gamma/3 — the continuum Dirac momentum operator — so E^2=m^2+|\mathbf p|^2 is recovered. Along a lattice axis, |S| = |\sin k|/3: it rises, peaks at k=\pi/2, and returns to zero at the zone boundary k=\pi. That return to zero is the doubler, stated honestly. The single-tick propagator’s eigenvalues tie the two sides together,
\lambda_\pm(\mathbf k)=i\sin(\omega/2)\;\pm\;\cos(\omega/2)\,|H_{\text{RGB}}(\mathbf k)|,
— the i\sin piece is the inertial channel, the \cos\cdot|H| piece is the propagating one. This is the same propagator whose directional content gives the lattice’s birefringence; that essay is about the angular shape of |H_{\text{RGB}}|, this one is about its magnitude and what it does to energy.
The picture
fourier_kernel: the lattice value |\sin k|/3 (green) tracks the continuum k/3 (gray dashed) at small k but peaks at k=\pi/2 and returns to zero at the zone boundary, where group velocity vanishes. C — dispersion. E(k)=\sqrt{m^2+p_{\text{eff}}^2} with rest mass m=\omega (the same \omega that set the split in panel A, since E(0)=m): the lattice dispersion curve (purple) bends below the continuum cone (gray dashed) and turns back toward the mass shell at the zone boundary — the continuum E^2=m^2+|\mathbf p|^2 is the low-energy limit. Panel B is the engine kernel directly; panel C combines it with E^2=m^2+p_{\text{eff}}^2 (the exact single-tick eigenvalue dispersion shares the same qualitative features — continuum recovery, bend-over, doubler).Why it matters
The lattice is its own ultraviolet regulator, for free. There is a maximum phase-rate per tick — phases separated by 2\pi are the same on a discrete clock — so the energy a single mode can carry is bounded by the tick rate, and the effective momentum is bounded by |S|_{\max}=1/3 (panel B). No cutoff is imposed by hand; it is the discreteness of the clock and the lattice. Continuum field theory pays for unbounded momenta with ultraviolet divergences and then regulates them after the fact. Here the bound is upstream of the dynamics.
Nothing in the substrate bends — there is nothing that could. The lattice is causal adjacency: which events neighbour which, and nothing more. The integer coordinates are bookkeeping for that adjacency and carry no metric of their own; distance, momentum, and energy are all emergent from the phase field on a fixed graph. So when the curve in panel C “bends,” it is a line on a derived energy–momentum plot, not a deformation of anything physical — the adjacency is untouched. This is the same reason the framework can have gravity without curved spacetime: what looks like bending — light deflected near a mass — is clock-density refraction, a gradient in the phase-advance rate V, the very V in \Delta\phi=\omega+V that sets the inertia split in panel A. One quantity, the phase-rate on a fixed adjacency graph, carries inertia, dispersion, and gravity — and none of the three deform the lattice. Conflating “the dispersion curve bends” with “the lattice bends” would quietly smuggle back the metric background the model is built to do without.
Inertia stops being mysterious. “Why does a massive thing resist acceleration?” becomes, on the lattice, “because a faster-rotating phase diverts more of its one amplitude into staying put.” Mass is the \sin^2(\Delta\phi/2) fraction; that is the whole mechanism, and panel A is the whole picture. It costs the model nothing extra, because it is the same amplitude that also does the propagating.
The continuum is recovered where it should be, and only there. Panels B and C agree with the relativistic forms at small k — that is the honest content of “deriving the Dirac equation.” They depart at high k, which is exactly where a real lattice should depart from a continuum idealisation, and where any falsifiable difference would live. In Paper I’s audit table this phenomenology already sits on three rows marked PASS: the dispersion-and-zone-boundary behavior of panels B/C is Photon dispersion (group velocity vs. wavenumber, lattice corrections at the Brillouin-zone boundary; exp_09), the inertia split of panel A is Inertial persistence (exp_01), and the small-k recovery is the Dirac equation row (derived analytically).
Open / next
What actually happens at the doubler? The return of |S| to zero at k=\pi is the fermion-doubling point, and this is where the strong form of the objection has to be answered, not waved away. The lattice is bipartite — RGB and CMY alternate by tick parity — which is the staggered-fermion family of mechanisms that lattice QCD uses to thin the doublers. Whether the bipartite tick rule disposes of the zone-boundary mode cleanly, or leaves a residue with physical consequences, is a specific, checkable question and is not settled here.
Is there a signature, not just a regulator? A regulator that is invisible is only half-interesting. The dispersion bends over at high k, so the lattice predicts an energy-dependent propagation speed — the same channel as the birefringence prediction (gamma-ray time-of -flight over cosmological baselines), but now in the magnitude of k rather than its direction. The question is whether the bend is large enough, at any accessible energy, to separate from the isotropic Lorentz-invariance bounds, or whether it lives only at the lattice scale where nothing yet reaches.
References
- Paper I — Geometry First: Quantum Mechanics, Gravity, and the Origin of the Standard Model from a Single Conservation Law. The bipartite tick rule, the \cos/\sin mass split (Zitterbewegung as the stay fraction), and the continuum Dirac limit live in the kinematics section; see the Paper I landing page.
- Companion essay — Birefringence in the A=1 lattice: the angular structure of the same propagator T(\mathbf k) whose magnitude structure this essay describes.
- Engine — the dispersion panels are computed from
dcl_core.core3d.HopOperator.fourier_kernel(the structure factor S(\mathbf k)) indcl-core; the figure script isnotes/phase_inertia_amplitude.py.