A neutral-atom quantum simulator has now resolved the finite-size excitation spectrum of an emergent conformal field theory and recovered its universal level ratios to within a few per cent. The instrument is general: it reads the spectrum of a system whose universality class is not known in advance. That makes it worth asking, precisely and without flattery, which quantities in this corpus such an instrument could confront. The answer is: fewer than the vocabulary suggests. The measurement tests ratios. Most of what dm³ asserts are scales, and a scale is invisible to a ratio.
This paper makes no claim that dm³ describes a Rydberg chain, no claim that the experiment discussed here bears on the correctness of anything in this series, and no claim of contact between contact geometry and conformal field theory. It is an exercise in falsifiability bookkeeping: given a real instrument with a known sensitivity, sort our own quantities into those it could in principle refute and those it could not, and state why. VALUE PREMISE A framework that cannot say which of its numbers are exposed to measurement has not finished being written.
DATA Sun, Le, Naus, Tsai, Picard, Murciano, Knap, Alicea and Endres [1] trapped chains of 7 to 35 ⁸⁸Sr atoms in optical tweezers and drove them to a Rydberg state, working in the blockade regime where adjacent atoms cannot both be excited. The effective model is Fendley–Sengupta–Sachdev [2]: a disordered phase, a charge-density-wave phase, and a second-order line between them that terminates at a tricritical point before turning first order.
The technique is modulation spectroscopy. Sit at the critical detuning, shake one Hamiltonian parameter sinusoidally at frequency f, then ramp into an ordered or disordered phase and count atoms. Population transfers when hf matches a gap between the ground state and an excited state, at a rate set by the transition matrix element. Sweep f and the peaks are the spectrum.
Ising point (central charge c = 1/2). Even-parity levels in ratio 2 : 4 : 6 : 8; odd-parity levels, reached by shaking antisymmetrically, in ratio 3 : 5 : 7. Reduced χ²/ν = 0.98 against a parameter-free prediction for L ≥ 19. Measured E₂/E₁ = 2.08(4) against 2.
Tricritical Ising point (c = 7/10). E₂/E₁ = 4/3 for free boundaries, 2 for fixed, 10/3 for the unstable boundary fixed point between them. Measured 1.45(25), 1.98(6) and 2.5(4). Boundaries are tuned by local detunings at the chain edges, with the unstable point at η ≈ 0.39.
Two features of that report are worth naming for reasons unrelated to physics. The L = 7 chain does not collapse onto the universal ratios, and the authors say so and explain it: at that size the second gap is comparable to the Rabi frequency, so the low-energy description does not apply. And 2.5(4) against 10/3 is described as "further" from prediction, attributed to a 1.7 % uncertainty in the next-nearest-neighbour interaction. The failures are stated by the authors before anyone else can state them.
The ladder 2 : 4 : 6 : 8 is not a numerological fact and should not be read as one. At the Ising transition the low-energy excitations are free Majorana fermions with linear dispersion. On an open chain of length L the allowed momenta are quantized,
and the boundary condition admits only an even number of fermions. A reflection-symmetric drive couples the ground state to pairs with |ka − kb| = π/L, whose total energies are 2, 4, 6, 8 in units of πħv/L. Shake antisymmetrically and the selection rule moves to 2π/L, giving 3, 5, 7. Linear dispersion, a box, and a selection rule. Nothing else is required, and the ladder is arithmetic in the plainest sense: En ∝ n.
At the tricritical point the fermions interact, no free-particle description survives, and the ratios stop being integers — 4/3, 2, 10/3 descend from the scaling dimensions of the relevant primary fields under each boundary condition. That is where the measurement stops being a check on kinematics and becomes a check on CFT data.
The word does at least three jobs in this series, and they have different exposure to measurement. Conflating them is the error this paper exists to prevent.
| Sense | Where | Object | What could bear on it |
|---|---|---|---|
| ρ-spectrum | Book 3 · Ch 11 Book 8 · Q.1 |
Eigenvalue moduli of a linear operator: ρ(A) = max|λᵢ|; the k-nacci roots ηk → 2 | Nothing empirical. These are theorems about roots of polynomials. |
| Flow spectrum | Vol II · §4 toy model |
Period T* of the limit cycle, transverse eigenvalue λ(z), Lyapunov exponent μmax | Relaxation and periodicity measurements on a system claimed to instantiate the flow. |
| Gap spectrum | not in this corpus | Many-body excitation energies above a ground state, and their ratios | Modulation spectroscopy, as in [1]. |
DATA Book 8 Q.1 already draws the first line itself. Its closing section — "What 'Ladder Toward ℵ₀' Means, and What It Doesn't" — states that the ηk sequence is a statement about polynomial roots and that the ladder language is structural analogy past that point. This paper extends that discipline outward rather than introducing it.
Book 3 Ch 11 asks its reader, in the chapter's own inquiry prompt, to name "the one observation or experiment that would prove this analogy wrong." Sections 4 and 5 below are an attempt to answer that prompt at research level rather than classroom level.
The reason the Ising measurement is a test at all is that v drops out. The light-cone velocity is non-universal — it depends on the lattice spacing, the Rabi frequency, the interaction strengths — and it is not predicted by the CFT. What the CFT predicts is the sequence of ratios, in which v cancels. A theory that predicted only v would be untestable by this instrument no matter how precisely v were measured, because the experiment would simply be calibrating it.
DATA This filter is not an invention of ours. It is the universal / non-universal distinction, which is the organising discipline of critical phenomena, and the paper applies it in its own vocabulary throughout: v is called "a non-universal velocity" wherever it appears, the level ratios are "universal energy ratios", and the L = 7 failure is attributed to "non-universal corrections". Universality is a theorem-grade statement, not a stance: under the renormalization group, microscopically unlike systems flow to the same fixed point and therefore share exponents and ratios, while every quantity carrying a scale — Tc, v, the lattice spacing — remains material-specific. The two-dimensional Ising class contains a magnet, a liquid–gas critical point, and now a chain of nineteen strontium atoms.
Stated as three clauses, which is the form this series should use, because the guard is then built into the claim rather than bolted onto it:
Scale-invariant — at the critical point, in the technical sense: correlations become power laws and no characteristic length survives. This is what lets a conformal description emerge at all.
Domain-agnostic — by universality, and for a stated reason: a universality class is fixed by symmetry and dimension, not by what the system is made of.
Predictive of the universal data only — ratios, exponents, central charges, degeneracy patterns. Silent by construction on every scale: where the transition sits, how fast the system moves, how large anything is. A framework of this kind that appears to predict a scale is either fitting it or has smuggled in a substrate.
A quantity is exposed to a spectrum measurement if and only if it survives multiplication of every energy in the problem by an arbitrary positive constant. Anything that does not survive is a scale: the measurement will fit it rather than test it, and reporting agreement is then reporting that a free parameter was fitted successfully.
| Quantity | Status under the filter | Exposed? |
|---|---|---|
| T* = 2π | Scale. A period, in units of time set by the vector field. Rescale time and it moves. | No |
| μ_max = −2 | Scale. A rate. Same rescaling argument. | No |
| τ = 2 | Scale — and worse. τ = √(c/κ_noise) equals |μ_max| in the toy model only because c = 2|μ_max| and κ_noise = 1. The second of those is a normalisation, not a derivation; the file that proves τ = 2 says so in its own docstring. | No |
| ε₀ = 1/3 | Ratio, but only because Γ = {r = 1}: ε₀ is a fraction of the cycle radius. Written as ε₀/rcycle it survives rescaling. | Yes, in principle |
| ε₀ = 1/3 ⟺ H = 2 | Ratio, and now kernel-checked. CLOSED 2026-08-26 eps0_eq_third_iff establishes that ε₀ = 1/3 picks out Hessian bound H = 2 uniquely among non-negative bounds, and H = 3 gives 1/4. | Yes |
| Γ₁₂ · 1:2 resonance | Ratio. A frequency ratio between the θ and z motions is dimensionless by construction and is exactly the kind of thing a spectrum resolves. | Yes — the sharpest |
| Neimark–Sacker | Ratio. A torus bifurcation predicts a pair of incommensurate frequencies — that is, the explicit absence of a rational ladder. MODEL, flagged pending the DNLS extension since 2026-08-07. | Yes — and it is the falsifiable one |
| η_k → 2 | Neither. A theorem about the dominant roots of λ^k − λ^{k−1} − ⋯ − 1. No measurement bears on it, and Book 8 Q.1 says so. | No |
| μ_max · T* | Ratio — the Floquet exponent over one period. Added 2026-08-28; the register below originally omitted it. μ_max = −2 and T* = 2π are each scales, but the product is not: rescale time by a and μ_max → μ_max/a while T* → aT*, leaving μ_max·T* = −4π fixed, i.e. Floquet multiplier e−4π. It satisfies the filter's own rule exactly. This is the gap spectrum — an eigenvalue separation rather than a polynomial root — and it is the standard observable for a limit cycle. | Yes |
The register above was built by testing each quantity separately against the rescaling rule.
That procedure is blind to dimensionless combinations: two quantities can each fail the
filter while a product or quotient of them passes it. μ_max · T* is exactly such a
combination, and its omission is a defect of method, not of arithmetic — every individual
row was decided correctly.
Two spectral objects were also being conflated, and they behave oppositely under the filter. The n-bonacci roots (φ, η, Δ, … → τ = 2) are dominant roots of a characteristic polynomial: algebraic constants, exposed to no measurement, and the register is right to say so. The gap spectrum — separations between eigenvalues, Floquet multipliers, ratios of rates — is dimensionless by construction and is precisely what a spectrum measurement resolves. The paper used "spectral" for both and inherited the wrong verdict for the second.
OPEN Biology is where this bites, and it is where this corpus has most of its instantiations. Circadian oscillation, cardiac dynamics, the autophagy cycle and neural rhythms are limit-cycle systems whose measurable invariant is the Floquet multiplier, not any polynomial root; increasing systemic complexity registers as a change in the gap structure. Those chapters were written before this filter existed and have not been re-read against it. Doing so is the work this correction opens, and the register should be expected to grow rather than shrink.
Three of nine, on the corrected register — and the third was found by asking what combinations survive rather than what quantities do. That is the honest count, and it is not a criticism of the framework so much as a description of what kind of framework it is: dm³ mostly fixes scales of a particular vector field, and a specified vector field is a model, not a prediction. The places where it makes a parameter-free statement are the stability radius, the resonance structure, and the Floquet gap.
OPEN For the resonance claims — the ones that pass the filter — the falsification has a shape, and stating it is the point of this paper:
Take a system independently argued to instantiate the dm³ flow. Drive one parameter sinusoidally, sweep the drive frequency, and record the response. If the framework's resonance structure is right, the response peaks stand in the predicted ratio, and that ratio is invariant under any change of the system's overall energy scale. If the peaks stand in a different ratio — or if the predicted 1:2 locking appears as an incommensurate pair, or the predicted incommensurate pair appears as a rational lock — the claim is wrong, and no adjustment of T*, μmax or τ can repair it, because those are the very quantities the ratio does not see.
The load-bearing premise is the first sentence and it is not supplied here: no system in this corpus has been independently argued to instantiate the dm³ flow to the standard that would make such a measurement interpretable. Until one is, the technique in [1] is a method we can describe and not a test we can run. Saying that plainly is worth more than an analogy.
VALUE PREMISE A resemblance between two numbers is evidence of nothing until a mechanism connects them. The paper discussed here is unusually rich in numerals that collide with ours, and every collision below is empty. They are listed so that a future reader of this corpus finds them already refused rather than freshly available.
| Theirs | Ours | Why it is a coincidence |
|---|---|---|
| E₂/E₁ = 2 (Ising) | τ = 2 | Theirs is a ratio of conformal descendant levels fixed by a selection rule; ours is √(c/κ_noise), a stochastic Lyapunov radius. No shared mechanism, and ours is a scale. |
| 2 : 4 : 6 : 8 | the g-series ladder | Theirs is arithmetic, En ∝ n, from momentum quantisation in a box. The k-nacci roots are algebraic and converge to 2 from below. A ladder is not the Ladder. |
| E₂/E₁ = 4/3 | ε₀ = 1/3 | Digit collision. Theirs comes from tricritical scaling dimensions under free boundary conditions; ours from |μ_max|/(2(1+H)) at H = 2. |
| ηc ≈ 0.39 | r* = 0.7759… | Unrelated quantities in unrelated systems. η is a dimensionless boundary-detuning strength; r* is a basin boundary. |
| c = 1/2, c = 7/10 | central-charge material | Central charges of two specific minimal models, fixed by the model. Nothing selects them from our side. |
OPEN The tricritical point is genuinely catastrophe-theoretic structure: it is where a Landau description needs a sixth-order term rather than a fourth. Volume II §5 asserts a correspondence between the four dm³ bifurcations and Whitney types labelled A₁ (fold), A₂ (cusp), A₃ (swallowtail). In Arnold's indexing — the one a referee from the singularity-theory or CFT side applies by default — Ak denotes the germ x^(k+1), so the fold is A₂, the cusp is A₃, and the butterfly is A₅. Under that convention the Ising point sits at the cusp and the tricritical point at the butterfly, and Volume II's "A₁–A₃" says something other than what it will be read as saying. The fix is cheap: declare the convention in the volume, once.
A second question is not merely notational. A Hopf bifurcation is a pair of complex-conjugate eigenvalues crossing the imaginary axis, which has no gradient-potential description and therefore does not appear in the Ak list at all. Volume II's defence is available — its correspondence is stated for singularities of the projection M = S × ℝ → S, not for critical points of a potential — but as printed the classification reads as catastrophe theory, and a reader will check it against catastrophe theory. OPEN until the projection argument is written out.
Criticality language is the most portable vocabulary in this series and therefore the easiest to misread. The misreading is always the same: a claim that two systems share a shape is heard as a claim that one supplies the mechanism for the other.
What is true, and what the experiment in [1] demonstrates twice: near a critical point, susceptibility to the parameters of the governing equations diverges. A local detuning at the two edges of the chain, tuned through ηc ≈ 0.39, decides which of two boundary universality classes the bulk spectrum lands in. A 1.7 % uncertainty in V₂ moves a predicted ratio from 3.33 to a measured 2.5(4). Small parameter, large qualitative consequence — and the tricritical point is precisely where the transition stops being reversible and becomes first order and hysteretic.
Two things travel with that, always. First, the sensitivity is to parameters, not to states. Unitary evolution preserves the overlap between two states exactly; there is no exponential state-to-state divergence of the classical-chaos kind. "Tiny changes" is true of parameter space and false of state space. Second, it is a property of sitting at criticality, not of being quantum, small or fundamental. Off the critical point susceptibility is finite and small perturbations do small things, which is why the world is mostly stable.
VALUE PREMISE This series does not claim, and its pages must not imply, that the quantum scale supplies the mechanism at the geological, biological or civilisational one. The shared structure is the threshold, not the scale. A quantity driven slowly to a critical value, a qualitative change when it is crossed, and often no way back — that shape is substrate-independent in the way a differential equation is, and it is the whole of what is being claimed.
This paper does not claim that a Rydberg chain realises the dm³ flow. It does not claim that conformal field theory and contact geometry describe the same objects. It does not claim that any measurement in [1] supports or undermines any result in this series — the experiment was designed to test conformal field theory and it tested conformal field theory. It does not propose an experiment we are in a position to run.
What it claims is narrower and, we hold, more useful: that of eight quantities this corpus treats as spectral, six are scales that any spectrum measurement would fit rather than test; that the two survivors are the stability radius and the resonance structure; and that the difference between those two categories was not previously written down anywhere in the series.
Counterpart to Book 3 · Ch 11 (ρ, spectral radius as the reach of unfolding) and Book 8 · Q.1 (spectral hierarchies, ηk → 2), which treat the ρ-sense. This paper treats the flow sense and the gap sense and separates all three. Method lineage: WP-29 (numerology sweep — no coefficient fitted to a dimensionless constant), WP-31 (calibration pipeline), WP-78 (what an audit cannot see). The register used here — sort your own quantities by exposure before claiming contact with an experiment — is WP-29's rule applied forward rather than retrospectively.
This paper contains no formal results. Its contribution is a partition of the series' own quantities into those a spectrum measurement could refute and those it could only fit, and a stated refutation condition for the two that survive.
sorryAx. A clean axiom report is not a reading of the statement: per R20, a theorem can assume its conclusion and still report clean. Follow the link before citing one as evidence.