Applications / Experiments read through FDT
Spin spirals wound by a helical density gradient
Piotr Put, Nathaniel Leitao, Haoyang Gao, Christina Spaegele and twelve colleagues wind nanometre-scale spin spirals into dense ensembles of nitrogen-vacancy centres in diamond with switched magnetic gradients that can reach up to approximately 2.2 mT/μm, read them out by winding back, and drive them between the two with Floquet sequences that make the dipolar interaction SU(2)-symmetric. A spiral wound and unwound with no quench in between revives the coherence at matched wave vectors; during a quench the spiral precesses, and the precession reverses sign when the winding turns from parallel to perpendicular to the selected crystal axis. FDT returns the measured record from the same source and data: general relativity is FDT in its spacetime regime and quantum mechanics is FDT in its matter regime, so agreement with the record is the expected outcome. The applied field is a macroscopic helical density gradient, and each step of the ensemble is one system-wide settlement. The analysis recomputes the record’s own factors. The angular average −j₂(ks)P₂(ξ𝐤) gives the reversal, with a parallel-to-perpendicular ratio of −2 and a zero at 54.74°; the radial integral’s small-wave-vector and large-volume limits differ; Eq. (1) prints a tensor of trace 8 where the dipolar tensor is traceless; the intermediate step of Eq. (12) is twice its endpoint; and with the stated spin definition the legend of the twisting fit prints half the classical slope of its one-axis-twisting Hamiltonian. Five proposed comparisons yield no prospective prediction, and the zero-bias torsion control is where FDT’s reading of a twist as a field meets the conventional time-reversal zero.
The record
What the article reports
The ensembles are nitrogen-vacancy (NV) centres in diamond, coupled by magnetic dipole–dipole interactions. A 532 nm green pump initializes the electronic ground-state spins and the red fluorescence is collected through the same objective in a confocal setup under ambient conditions; repeated resonance scans align the bias field to the selected crystallographic axis η̂, and the two levels used are the magnetic sublevels 0 and −1 of the ground-state triplet. In the bulk sample’s confocal spot the single-group NV density is estimated at 246 ppb from an XY16 decay timescale of 21.5 μs, through a simulation-based conversion that assumes the decay is dominated by dipolar interaction; the typical spacing is about 28 nm, the doped layer is 185 nm thick, and the sample sits on a two-wire chip whose current ratio tunes the gradient. The nanobeam, a triangular beam of approximately 3.8 ppm per group and a typical spacing of about 11 nm, sits on a microcoil run with a pulsed current. The dipolar scale Jdip = 2πħ × 52 MHz nm³ sets the characteristic exchange strength. The gradients can reach up to approximately 2.2 mT/μm, and at this gradient strength the typical Zeeman shift difference between neighbouring spins separated along the gradient exceeds their dipolar interaction by at least one order of magnitude.
A global microwave rotation sets the spins on a cone of angle ϑp. A gradient in two equal halves of opposite polarity, with a microwave half-turn between them, then winds their azimuths along the gradient while the static detunings cancel. After a quench under periodic driving, an inverse winding and a global quadrature measurement select one Fourier component of the transverse spin. Winding and unwinding with no quench in between, the authors observe a clear revival in the coherence magnitude centred at matched winding and unwinding, which they describe as evidence of reversible dynamics under the gradient field: for the equatorial bulk spiral of pitch 0.242 μm the magnitude revives near matched wave vectors, with an approximate width of 2π/(0.4 μm) and a Gaussian fit. Two antipodal pairs suppress spurious global rotations, and Figs. 2(g)–2(i) plot the antipode-averaged imaginary Fourier component.
The native uniform-mode control in bulk uses XY-16-type decoupling, and the SU(2)-symmetric interaction is engineered with cXY8-DROID-vXY4-Mirror in bulk and cXY4-DROID-vXY4-Sym in the beam. Without a spiral the antipode-averaged signal shows no observed time dependence; at finite wave vector the traces evolve. Changing the current ratio on the bulk wires, the authors observe opposite directions of precession for parallel and perpendicular propagation at fixed preparation, which is distinct from changing the hemisphere of the initial spins; in their Fig. 3(k) the abscissa is the angle between the winding direction and the crystal axis, at a magnitude of 2π/(0.242 μm). The native interaction gives the faster early-time precession and the engineered one eventually the larger amplitude, which the authors attribute to a Floquet rescaling of the exchange term by 1/3 and to the conservation of the summed transverse spin components that protects the spiral order from local relaxation.
In bulk, the fitted early-time precession rate rises rapidly, then saturates and eventually decays with wave vector, with a marker at 2π/(0.185 μm), and a Fourier image along a different axis shows the spins confined axially by the 185 nm sample thickness, consistent with a rectangular-profile fit to the raw data. In the beam, the maximum normalized amplitude of NV group one carries a width marker at 2π/(0.3 μm); the record gives no reference amplitude for that ordinate’s normalization, and its solid curve imports the measured winding loss. A Fourier image across the beam’s short axis resolves a dip that the authors state arises from nanophotonic interference of the pump light, confirmed by electric-field simulations, and simulated cross sections show two lobes for short pumping that merge at longer pumping.
In Fig. 5(a) spin tomography after unwinding follows the spiral through quenches up to 289.44 μs. With improved drive and gradient homogeneity at 2π/(0.242 μm), the precession rate extracted by linear fits exhibits a cosine dependence on the cone angle, which the authors call characteristic of one-axis-twisting dynamics, and the observations closely resemble that Hamiltonian’s dynamics with an effective twisting rate of 8.51(2) kHz. At a cone angle of 45° the spiral amplitude decays markedly more slowly at longer wavelength, with stretched-exponential fits, and the quality factor ω/γ, the precession rate over the fitted decay rate, is below the theoretical maximum of √(3/2) ≈ 1.22 for dilute, positionally disordered two-dimensional Ising spins at unity polarization for the native interaction, while for the engineered interaction the observations clearly exceed it for a broad range of wave vectors. In the authors’ outlook, with long coherence times and high-fidelity readout, entanglement generation and spin squeezing become feasible, and a cited analysis predicts that in isolated, disordered dipolar systems such spiral-mediated twisting can, in principle, achieve an ideal O(√N) quality factor, owing to the separation of timescales between spiral relaxation and exchange-driven twisting.
| Quantity | Value | Its verb |
|---|---|---|
| Dipolar scale | Jdip = 2πħ × 52 MHz nm³ | sets the characteristic exchange strength |
| Gradient | up to approximately 2.2 mT/μm | can reach |
| Bulk NV density; spacing | 246 ppb per group in the confocal spot, from a 21.5 μs XY16 decay; about 28 nm | is estimated; typical spacing is |
| Beam NV density; spacing | approximately 3.8 ppm per group; about 11 nm | characterized in earlier works; with a typical spacing |
| Bulk layer thickness | 185 nm | observe NVs confined axially by; consistent with a fit |
| Equatorial bulk spiral | pitch 0.242 μm; revival width about 2π/(0.4 μm) | prepare; the finite width reflects the extent of the polarized region |
| Direction scan, Fig. 3(k) | |𝐤| = 2π/(0.242 μm) | showing tunable strength and sign of the exchange field |
| Finite-size markers, Fig. 4 | 2π/(0.185 μm) in bulk; 2π/(0.3 μm) in the beam | saturation occurs around; vertical dashed line |
| Bulk Floquet sequence | 40 ns half-turn, 10 ns spacing, 50 ns padding per gradient edge | uses; padding is applied |
| Beam Floquet sequence | 30 ns half-turn, 30 ns spacing, 10 ns padding per gradient edge | uses; padding is applied |
| Field samples, Fig. 1(e) | a 200-nm-radius region | should be understood as the average within |
| Simulated spins | 1572 in a bulk cylinder 500 nm across and 185 nm high; 6750 in the beam | simulate |
| Ultraviolet cutoff | 0.2 times the spacing: 6 nm in bulk, 2.2 nm in the beam | is assumed to be |
| Imaging resolution | below 20 nm | high spatial resolution, in this work |
| Longest quench, Fig. 5(a) | 289.44 μs | marks |
| Effective twisting rate | 8.51(2) kHz | closely resemble one-axis twisting, with |
| Two-dimensional Ising maximum | √(3/2) ≈ 1.22 | indicates the theoretical maximum |
The FDT reading
P-5 the field, 0.1a one settlement, I-4 the quantum face
There is one substance, the omnium, and the chain E/m = c² = 1/(ε₀μ₀) = (d/t)² is one state read four ways, with c the strand rate (0.1a, 0.1d). Every step is one system-wide, many-to-many settlement of the state the system is in, and events do not commute (0.1a). General relativity is FDT in its spacetime regime and quantum mechanics is FDT in its matter regime, so wherever either holds FDT returns its results, and it departs only where it supplies an input the standard account lacks.
A magnetic field is a macroscopic helical density gradient of the omnium (P-5). The coil’s gradient along the selected axis is what winds the spiral: with 𝐠B the spatial gradient of the projected coil field, ytog the microwave toggling sign and gNV the winding coefficient in angular frequency per tesla,
in radians per metre. A pattern whose rest energy depends on where it sits in a twist takes that energy as its clock factor, the helical window (0.1c), with E₀ the rest energy outside the twist, Eind the induced-moment energy with its induction work, 𝗆j and 𝐁j the held moment and the twist field at site j, and 𝓜 the body’s rest mass:
so a uniform twist turns a held moment and pushes nothing, preparations whose net projected energy gradients per whole-body rest energy differ accelerate differently along the gradient, and gravitational monopole free fall stays universal. On the quantum face a projected helix’s phase turns as a rotor, eℬφ = cos φ + ℬ sin φ with ℬ the unit bivector of its plane, and phase ignorance at observation is superposition (I-4). The detector’s two quadratures are the two components of the weighted sum of the sites’ projected helices, one complex number:
with ϖj each site’s real weighted transverse amplitude and 𝐱j its position. At equal winding and unwinding profiles the spatial phases cancel, and a common phase rotates the two quadratures without changing the magnitude. The two quadratures are distinct components of one number, and no single microscopic phase can be reconstructed from it.
After taking out a positive prefactor, the record’s Eq. (1) prints the tensor 3𝖨 − 𝐝̂𝐝̂T, with 𝐝̂ a pair’s direction. Its trace is 8, while the magnetic dipolar tensor 𝖨 − 3𝐝̂𝐝̂T is traceless, and no overall scalar removes the difference, so the comparison starts at the record’s independently stated Hamiltonians. With ξ = 𝐝̂·η̂ and P₂(ξ) = (3ξ² − 1)/2,
Every pair scalar product of Hiso commutes with each collective spin component, and a fully aligned product is a triplet on each pair, stationary up to phase; at finite wave vector the position phases weight the two commutators of a pair differently. The native Hamiltonian conserves the axial collective component and its angular factor integrates to zero with an isotropic weight; neither fact sets every site’s transverse torque to zero in a finite disk.
The directional dependence of the precession is a degree-two harmonic. Over the sphere of pair displacements of length s, a plane wave against the traceless angular factor keeps only its degree-two part, so with ξ𝐤 = 𝐤̂·η̂ and j1, j2 the spherical Bessel functions,
and the filter 1 − cos(𝐤·𝐝) of the record’s Eq. (9) carries +j₂(ks)P₂(ξ𝐤), its radial weight j₂. The angular factor is 1 for parallel winding and −1/2 for perpendicular winding, a signed parallel-to-perpendicular ratio of −2, with its zero at arccos(1/√3) = 54.74°: the reversal the record observes. The radial part, with xlo = kℓlower and xhi = kℓpol for the lower cutoff and the sphere’s radius, is
The primitive j₁(x)/x tends to 1/3 at zero and falls as the inverse square of its argument, so for xlo ≪ 1 ≪ xhi the integral approaches 1/3, while at small k and fixed cutoffs it vanishes as k². The record’s Eq. (8) starts the radius at zero and its Eq. (10) introduces a positive lower cutoff; replacing the upper limit by infinity before taking small k changes the answer. With uniform density ρuniform the exchange energy is Jsp = 4πρuniformJang(ξ𝐤)Irad/2, and the torque and polarization factors enter the measured slope after division by ħ.
The contact term checks the normalization independently. Taken as a spherical principal value at the origin, P₂(𝐝̂·η̂)/s³ extends to half of (η̂·∇)²(1/s) + (4π/3)δ⁽³⁾(𝐝): away from the origin the axial second derivative of 1/s is 2/s³ while P₂(1)/s³ = 1/s³, and the trace at the origin is the Coulomb Laplacian. The half-normalized transform at nonzero k is −4πP₂(ξ𝐤)/3. The record’s Eq. (11) omits the common half, and the intermediate expression in its Eq. (12) is twice its printed endpoint; the directional reversal survives the check.
The sign of the turn and the order of limits. Direction: the angular factor of the precession, P₂(ξ𝐤), against the angle between the winding vector 𝐤 and the selected crystal axis η̂, the abscissa of the record’s Fig. 3(k); the slider sets the angle, and the inset draws the two directions. The factor is 1 for parallel winding and −1/2 for perpendicular winding, so the precession reverses with a signed parallel-to-perpendicular ratio of −2, and it vanishes at arccos(1/√3) = 54.74°; the measured slope also carries the radial integral, the torque and the polarization, which the factor multiplies. The order of limits: the radial integral Irad against xhi = kℓpol on a logarithmic axis, for a sphere whose radius is ℓpol/ℓlower times the lower cutoff, set by the slider; the grey dashed lines are the small-k series (xhi² − xlo²)/30 and the value 1/3, and the coloured dashed curve takes the upper limit to infinity first, j₁(xlo)/xlo, which tends to 1/3 at small k where the finite sphere’s integral vanishes. Both views are dimensionless and use no value of the record beyond the angles.
I-4, Use 2
What the readout fixes
The twisting slope, its units and the beam’s free axis
For N effective spin-half sites the Bloch polarization is pz = 2⟨𝗌collz⟩/N, the collective component being the sum of the site components. The record’s one-axis-twisting Hamiltonian, in angular-frequency units, and the classical transverse phase slope that follows by differentiating it with respect to the collective axial component are
with pz = cos ϑp for spins fully polarized at preparation, which is the cosine the fitted slopes exhibit. The legend of the record’s Fig. 5(c) multiplies the same parameter by ⟨𝗌collz⟩/N: with the stated spin definition that expression is half the classical slope of the printed Hamiltonian, so preserving both the fitted parameter and the printed curve requires stating which spin normalization is plotted. Fig. 5(b) plots phase in degrees against quench time in microseconds, and its slopes put the ordinate of Fig. 5(c) in thousands of radians per second: the reported 8.51(2) kHz is an angular rate, and its cyclic value is that rate divided by 2π. Refitting the points or assigning full confidence coverage to individual slopes needs the complete fit windows and raw replicate covariances, which the record leaves unspecified.
The twisting fit concerns a mean phase. A first moment gives no spin variance, so readout variance and metrological gain require further observables: independent spin variance, contrast, detector-noise calibration and a phase-sensitivity reference. The quality factor divides two fitted rates, and at first order its variance is Var[erate − (ωsp/Γsp)edecay]/Γsp², with Γsp the e-folding rate of the amplitude envelope and erate and edecay the two fit fluctuations, so their covariance enters it. The reported scan varies the wave vector on a fixed specimen, while the cited expectation concerns larger systems.
The beam’s Fourier image across its short axis is a marginal: it integrates over the polarization along the beam’s long axis, so profiles with the same measured marginal can differ along the beam. The analysis bounds how far that freedom moves the exchange response, as an exploratory in-sample calculation (X-1). Along the long direction it places either one slab or two translated copies with equal shares about the same midpoint, each carrying the same normalized transverse column distribution, with centre separation ℓsplit, slab thickness hslab and pair exclusion length ruv under the illustrative restrictions, with Wbeam the beam’s largest transverse diameter, Wbeam/ℓsplit ≤ 1/4, ruv/ℓsplit ≤ 1/4 and hslab ≤ ruv²/(100ℓsplit), and takes the wave vector and quantization axis along the separation with kℓsplit = π. In units of the angular energy–volume prefactor at parallel alignment over ℓsplit³, per contributing spin, the kernel is
with Δz the axial pair displacement, s the pair distance, Bin the same-slab integral and Bcross the cross-slab average. The one-slab result is Bin and the two-slab result (Bin + Bcross)/2, so the spread is exactly (Bcross − Bin)/2 for every allowed marginal. Using 1 − cos ζ ≤ ζ²/2 bounds |Bin| by π²ℓsplithslab²/(4ruv³); for cross pairs |Δz|/ℓsplit lies between 0.99 and 1.01, the angular cosine squared is at least 0.99²/(0.99² + 0.25²), the phase filter at least (1 + cos 0.01π)/2 and s/ℓsplit at most √(1.01² + 0.25²), and the cross kernel is at most 1/0.99³. Combined and rounded outward, these place the two-slab response minus the one-slab response between 0.40 and 0.52 in the illustrative units, for every admissible column profile in the family. The span is neither an experimental uncertainty nor a force; the absolute energy scale needs the actual spin density and collection calibration, and a held-out image along the beam constrains the separation and the shares through its own transfer function and error covariance.
The twisting slope and the free axis. The twisting slope: the transverse phase in degrees against quench time, over the maximum quench time of 289.44 μs in the record’s Fig. 5(a), at the cone angle the slider sets. The solid line is the classical slope χtwpz with pz = cos ϑp and the reported 8.51 read, as Fig. 5(b) requires, in thousands of radians per second; the dashed line is the legend’s expression with the stated spin definition, half of it; the dotted line reads the same number in thousands of cycles per second, 2π times the solid one. The lines are the formulas applied to the reported parameter, not the record’s data points. The free axis: a schematic of the beam, not its measured profile, with one slab or two translated slabs along its long axis; both share the measured short-axis marginal, which integrates over that axis. The slider sets the beam’s width over the slab separation, up to the analysis’s 1/4, and the bars show the bound on the cross-slab kernel and on the two-slab minus one-slab response that the restrictions give at that width; at 1/4 the response lies between 0.40 and 0.52, the analysis’s interval, in units of the angular energy–volume prefactor at parallel alignment over ℓsplit³.
X-1, Use 2
What a discriminating test needs
Five comparisons, none counted as a prediction
A divergence is claimed only where FDT supplies an additional source, a constitutive input or a changed observable dictionary. Each proposed comparison reads an active setting against a reference with independently calibrated conventional forecasts for both, and with 𝐲 the vector of the active reading, the reference reading and their two forecasts, in that order, and Σ their joint covariance,
In a calibrated linear range with jointly Gaussian errors and known covariance the exact two-sided size of each test is erfc(3/√2) = 0.0027; shared drift, correlations between regions and forecast uncertainties enter the contraction, phases need an independently established unwrapped range, and selecting the largest of five discrepancies needs its joint selection law. In each entry below the four numbers are the residual mean with the FDT input, the residual mean with its conventional value, their difference and the threshold, in units of the standard uncertainty.
- DC-1Static torsion and magnetic phase: a twisted beam with two mapped NV regions. Each torsional polarity is held at rest and both are taken in random order; since an equal-half echo cancels a static field’s phase at each site, a separately referenced DC-sensitive Ramsey interval reads the static field, and holding the total field fixed fixes that phase in both calculations. The comparison is retired for the shared projected Zeeman response, (0, 0, 0, 3) (SA-01). A zero-bias control nulls the external bias at a target with no selected magnetic order: in conventional time-reversal-invariant equilibrium static strain is even and magnetization odd, so the equilibrium magnetization vanishes for either polarity. P-5 reads a static twist as a magnetic field, so at zero external bias FDT places a torsion-odd field where the conventional equilibrium places none; the twist-to-field kernel that sizes it is to be derived (D-01), and the comparison holds no FDT residual mean to test.
- DC-2Order contrast with an open spin system: two orders of the same gradient blocks. Two permutations of labelled collinear gradient-area blocks, matched in duration and site-resolved phase, are read on the same referenced quadrature, each order’s ordinary open-system propagator and readout computed with its transients, bath history and attenuation. Equal response maps give equal order contrasts even for a finite ordinary order dependence. Retired for the shared ordered spin and readout map, (0, 0, 0, 3) (SA-02): each order is its own system-wide settlement, and the core names no objective that adds a history-to-quadrature term (0.1a).
- DC-3Exchanged preparation on elastic supports: two sensors, the populations swapped. The zero and minus-one populations are exchanged between two matched sensors in mapped gradients, and the change in static deflection is the exchanged-state difference of the incremental magnetic force at fixed populations over the calibrated stiffness. Retired for the matched rest-energy and support calculation, (0, 0, 0, 3), including a finite preparation-dependent deflection (SA-03): at one energy the helical window returns the conventional gradient force, and inertia is the rest energy of the whole moving body over c² in both descriptions.
- DC-4Loaded gap: a Ramsey frequency in a loaded region against its neighbour. The observable per region is the clock ratio times the level-energy difference over h, in hertz, and a shared additive energy cancels from each gap. Retired for a stipulated common additive level shift alone, (0, 0, 0, 3) (SA-04): the full load response carries the two level sensitivities and the local clock-ratio response, and no form of the confinement operator or its spectrum enters here (O-9).
- DC-5Optical path: a weak interferometric probe across loaded and reference diamond paths. The signed phase is the active-minus-reference difference of the real propagation-constant integral plus the interface phases, with both calculations carrying electrodynamics in diamond and the same metric and boundaries. Retired for the common material phase integral and interfaces, (0, 0, 0, 3) (SA-05): the aperture index is an index of the gravitational optical metric that both calculations carry (O-8), so it cancels from the loaded-diamond contrast.
In each comparison every contribution that FDT and the conventional calculation both specify takes one value in both; the torsion field that FDT assigns to a static twist, the load response of the defect levels and the causal aperture enter no number here. None is counted as a prediction. The zero-bias torsion control is where FDT’s reading of a twist as a field meets the conventional time-reversal zero, and the twist-to-field kernel turns that reading into a signed differential prediction in its own units.
Readings declined
Use 6 Data yes, interpretation not necessarily
Standard reading
A clear revival in the coherence magnitude centred at matched winding and unwinding, without an intermediate quench, provides evidence of reversible dynamics under the gradient field.
FDT reading
The measured refocusing is accepted in full. The unwinding is a new settlement of the whole state it starts from, which restores the site phases without running the winding’s microscopic events and environmental records backward; events do not commute, and a process run backward is never the mirror of the forward event (0.1a).
Quantum spin-spin interactions are approximated by classical magnetic fields acting on each spin, and the observed precessions are driven by the dipolar exchange field because angular averaging removes the Ising contribution in a three-dimensional dipolar ensemble.
The precession of the detected mode is the settlement of the whole ensemble at every step (0.1a), with the applied field a macroscopic helical density gradient (P-5), the site-field approximation the consensus reduction of the ensemble propagator, and the quadratures the two components of the weighted sum of the sites’ projected helices (I-4). The measured rates are accepted in full as the record of that settlement.
The standard account reads the transverse spin coherence as a probability amplitude over the two sublevels.
The transverse components are the two components of a projected helix’s phase rotor, and superposition is the ignorance of that phase at observation (I-4). The universe is deterministic; the coherence and its revival are accepted in full as ensemble readings of the weighted phases.
Convergence
Use 2 Same source, same data
At the same source, pulse waveforms, stress tensor and boundary data, the propagator of the whole ensemble under the selected sum of pair couplings, with its ordinary open-system channels, followed by inverse winding and the calibrated optical projection, returns the complex spin coherence and the same fitted rates as the consensus calculation, and supplying a measured ancillary decay to both returns the same empirically corrected curve. None of these operations takes the curvature ratio ασ, a chart or a branch tag as an input. FDT recovers every reading of the experiment from the same source and data: the refocusing, the directional reversal, the finite-size behaviour of the fitted rate and the cosine law of the twisting slopes.
FDT makes a divergence claim only where it supplies an additional source, a constitutive input or a changed observable dictionary. In the proposed modified devices ordinary magnetic, elastic, spectroscopic and optical responses already survive the comparison, and the torsion field of a static twist, the load response of the defect levels and the causal aperture enter no number here (D-01, O-9, O-8). Where FDT, general relativity and quantum mechanics converge, FDT reproduces their results; where it declines their assumptions, here a field as an entity apart from the omnium, a refocusing that runs events backward and probability amplitudes, it is under no obligation to reproduce them.