Precision AMO Frontiers

Status: coherent quantum transduction, atomic clocks, atomic magnetometers, and atom interferometers are established. Extending their performance across networks, long baselines, moving platforms, and entanglement-assisted protocols is active. Sensitivity projections for undemonstrated instruments are conjectural. An AMO anomaly attributed to a new particle or force remains conjectural until conventional models, global statistical significance, and independent measurements have been tested.

Last reviewed: 26 July 2026. Record values and project milestones below are date-stamped. They are not permanent platform rankings.

Core Question

How much reliable information about time, fields, motion, constants, and new interactions can an AMO experiment extract per unit resource while preserving traceability and control of systematic error?

The phrase “per unit resource” matters. A sensor can improve its single-shot phase resolution while taking longer to prepare, losing duty cycle, narrowing dynamic range, or becoming more sensitive to an uncontrolled parameter. A frontier claim should therefore specify at least:

• the measurand or searched-for coupling;
• the response function and accepted signal band;
• statistical uncertainty per shot and per wall-clock time;
• systematic and model uncertainty;
• spatial resolution, bandwidth, dynamic range, and dead time;
• preparation, calibration, and analysis overhead;
• environmental operating conditions; and
• the comparison baseline.

The frontier is not one race toward smaller units. It is the search for better validated measurement systems.

Why It Matters

Precision AMO systems serve three related but distinct roles.

Metrology

They realize and compare units, most prominently frequency and time. Optical clocks now resolve shifts far below the uncertainty of the caesium realization that presently defines the SI second. The open problem is no longer only how to build one excellent clock. It is how to compare, disseminate, operate, and reproduce optical references internationally.

Sensing

Quantum states transduce magnetic fields, electric fields, acceleration, rotation, gravity gradients, temperature, and electromagnetic radiation into frequency, phase, or population. The central research problem is to preserve laboratory-grade calibration under realistic motion, drift, gradients, limited power, and finite volume.

Fundamental tests

Clock ratios, spin precession, atomic recoil, molecular internal fields, and matter-wave phases probe dimensionless constants, discrete symmetries, equivalence principles, possible dark-matter fields, and new forces. These experiments can constrain interactions at energy or length scales not directly accessible to colliders. The translation from a null observable to a particle model is theory dependent and must remain explicit.

Canonical Boundary

This page owns the dated cross-platform frontier assessment. It does not repeat the mature derivations.

• Precision Measurement and Metrology owns measurands, traceability, Allan statistics, uncertainty propagation, projection noise, and the full AMO measurement chain.
• Optical Clocks owns ion and lattice architectures, frequency-comb readout, blackbody and lattice shifts, clock comparisons, and relativistic geodesy.
• Atom-Interferometric Sensors owns sensitivity functions, inertial scale factors, gravimetry, gradiometry, gyroscopy, phase unwrapping, and sensor-level corrections.
• Magnetometry owns optical pumping, Zeeman transduction, spin-exchange relaxation-free operation, transfer functions, heading error, and array calibration.
• Fundamental Constants owns correlated least-squares adjustment, atom-recoil determinations of the fine-structure constant, mass ratios, Rydberg-scale quantities, and magnetic moments.
• Variation of Constants Searches owns sensitivity coefficients, drift, periodic and transient signal models, spectral windows, and network covariance.
• Tests of Fundamental Symmetries owns EDM and parity observables, switch-parity analysis, molecular enhancement, theory response matrices, and model-dependent inference.

Optical Clock Frontiers assesses clock-specific research, while Fundamental Symmetry Frontiers owns the dated cross-program ledger for EDMs, parity violation, nuclear moments, and new bosons. This page compares the precision ecosystem as a whole.

Minimal Background

A unified response model

Most precision sensors can be written as a response plus nuisance terms. In the time domain,

$$y_i(t) = \int_{-\infty}^{\infty} h_{i\theta}(t-t') \theta(t')\,dt' + \sum_j \int_{-\infty}^{\infty} h_{ij}(t-t') x_j(t')\,dt' + n_i(t).$$

Here $y_i$ is a recorded channel, $\theta$ is the measurand or searched-for signal, $x_j$ are environmental or control inputs, $h_{i\theta}$ and $h_{ij}$ are calibrated impulse responses, and $n_i$ is residual noise. In the frequency domain,

$$\widetilde y_i(f) = H_{i\theta}(f)\widetilde\theta(f) + \sum_j H_{ij}(f)\widetilde x_j(f) + \widetilde n_i(f).$$

This form covers apparently different devices:

Platform	Recorded coordinate	Leading response
clock	excitation probability or servo correction	detuning integrated by the interrogation sensitivity function
atomic magnetometer	optical rotation, transmission, or spin projection	$\omega_L=\gamma B$ within a calibrated response band
light-pulse atom interferometer	output population or differential phase	$\phi_a\simeq\mathbf k_{\mathrm{eff}}\cdot\mathbf a\,T^2$ for the ideal three-pulse sequence
constant comparison	frequency ratio or recoil phase	logarithmic sensitivity to dimensionless constants
new-force search	phase, frequency, displacement, or torque residual	model-specific spatial, temporal, and composition response

The compact equations are scale factors, not full instruments. The transfer function, sign convention, spatial averaging, and nuisance channels decide what the recorded number means.

Information per wall-clock time

The quantum Fisher information $F_Q$ bounds the variance of an unbiased single-parameter estimator under its state model,

$$\operatorname{Var}(\hat\theta) \ge \frac{1}{\nu F_Q}.$$

If each experimental cycle takes $T_c$, a useful idealized information rate is

$$\dot F_Q = \frac{F_Q}{T_c}.$$

State preparation, verification, adaptive control, rejected shots, and recalibration all contribute to $T_c$. A protocol with larger $F_Q$ but much longer cycle time may have a lower information rate. Neither expression contains unknown systematic bias or model discrepancy.

Three uncertainty layers

The frontier requires three separate ledgers:

$$u_{\mathrm{total}}^2 = u_{\mathrm{stat}}^2 + u_{\mathrm{syst}}^2 + u_{\mathrm{model}}^2.$$

This schematic sum is exact only when the three groups are independent and represented by standard uncertainties. In real analyses, covariance and nonnormal likelihoods may be important. The conceptual separation remains:

• statistical uncertainty concerns finite and noisy records;
• systematic uncertainty concerns calibrated influences on the measurement equation; and
• model uncertainty concerns omitted physics or an imperfect translation from observable to target parameter.

What Is Known

Optical clocks have crossed the $10^{-18}$ frontier

Optical lattice and trapped-ion clocks have demonstrated systematic uncertainties at or below the low-$10^{-18}$ scale. A 2025 $^{27}\mathrm{Al}^{+}$ quantum-logic clock reported a fractional systematic uncertainty of $5.5\times10^{-19}$ together with improved ion-clock stability. This is an established, peer-reviewed instrument result for the stated clock and uncertainty model.

It does not mean that global time transfer, every optical species, or routine clock operation has the same uncertainty. The performance chain is

$$\begin{gathered} \text{local clock uncertainty} \longrightarrow \text{ratio comparison} \longrightarrow \text{link uncertainty} \\ \longrightarrow \text{interlaboratory agreement} \longrightarrow \text{time-scale contribution} \longrightarrow \text{user access}. \end{gathered}$$

As of this review, the SI second is still defined by the unperturbed $^{133}\mathrm{Cs}$ ground-state hyperfine frequency. The BIPM states that presentation and consideration of an optical redefinition could occur from 2026, while ratification is possible no earlier than 2030. The date and choice of definition remain metrological decisions subject to mandatory criteria.

Clock ratios are becoming the primary consistency test

Absolute optical frequencies inherit the caesium realization. Optical ratios compare high-performance clocks more directly and form closure tests across species. For three clocks $A$, $B$, and $C$, ideal ratio closure requires

$$\ln R_{AB} + \ln R_{BC} + \ln R_{CA} = 0.$$

In April 2026, a NIST-hosted preprint reported $^{27}\mathrm{Al}^{+}$, $^{171}\mathrm{Yb}$, and $^{87}\mathrm{Sr}$ ratios with total fractional uncertainties at or below $3.2\times10^{-18}$ and stated that the measurements satisfy a redefinition milestone criterion. The same report noted discrepancies with earlier ratios at larger fractional levels. The correct frontier interpretation is:

• established: cross-species ratios can be measured at the few-$10^{-18}$ level in that campaign;
• active: reconcile historical discrepancies and establish independent, geographically distributed consistency;
• not established: that one network by itself settles the definition or identifies the source of every discrepancy.

The preprint status should be retained until a version of record appears.

Atomic magnetometry is a mature transducer with active deployment limits

Optically pumped magnetometers convert spin precession to an optical or electrical signal. Spin-exchange relaxation-free operation, driven resonances, free precession, gradiometric arrays, and RF sensing are established architectures. Applications include biomagnetism, low-field NMR, field mapping, navigation, materials characterization, and searches for spin-dependent interactions.

The frontier is not whether atoms respond to magnetic fields. It is whether the sensor retains a calibrated vector response in the presence of:

• field gradients and motion;
• finite shielding and Johnson noise;
• light shifts and optical backaction;
• heading and dead-zone errors;
• sensor-to-sensor crosstalk;
• bandwidth–sensitivity tradeoffs;
• biological or geophysical backgrounds; and
• source localization uncertainty.

A quoted amplitude spectral density $S_B^{1/2}(f)$ is a noise property in a declared band. It is not automatically the uncertainty of a static field, the detection limit for an unknown waveform, or the resolution of an array.

Atom interferometers measure inertial phase in controlled regimes

Light-pulse atom interferometry is established for acceleration, gravity, gravity gradients, and rotation. Laboratory gravimeters and gradiometers can operate with well-tested scale factors and detailed corrections. Portable systems and hybrid classical–quantum inertial sensors have also been demonstrated.

Two recent results illustrate distinct active directions:

• In 2025, a two-dimensional Bose–Einstein-condensate array demonstrated simultaneous sensitivity to acceleration, rotation-related mirror motion, gravity gradients, and higher spatial derivatives. This is an established multi-axis proof of principle, not yet a universal navigation solution.
• In June 2026, an AION prototype used two separated $^{87}\mathrm{Sr}$ clock-transition interferometers interrogated by a common laser. Under several radians of synthetic laser phase noise, its differential estimator was consistent with its projection-noise model and recovered coherent injected signals. This is an established prototype milestone for common-mode rejection. Long baselines, larger atom number, wavefront propagation, large momentum transfer, and full detector sensitivity remain active.

No gravitational wave or dark-matter detection follows from the prototype. It validates an ingredient of proposed detectors.

Fundamental constants are network-adjusted quantities

The current internationally recommended values are the 2022 CODATA adjustment, released on the NIST constants database in 2024 and published as a full adjustment report in 2025. The database notes that the next regular adjustment is the 2026 CODATA cycle.

Some SI defining constants, including $h$, $e$, $k$, and $N_A$, have exact assigned numerical values. Measured constants such as $\alpha$, particle masses, and $G$ retain uncertainty. A new observation does not immediately replace the adjusted value.

The fine-structure constant illustrates why independent routes matter. Atom recoil relates

$$\alpha^2 = \frac{2R_\infty}{c} \frac{A_r(X)}{A_r(e)} \frac{h}{m_X},$$

for atomic species $X$, with correlated auxiliary quantities. A 2020 rubidium recoil determination reached $81$ parts per trillion and differed by more than five reported standard deviations from the then-leading caesium recoil result. That tension motivates metrological investigation. It does not, by itself, establish new physics: recoil systematics, input covariance, and comparison with the independently measured electron magnetic anomaly all enter.

Precision AMO has constrained, but not discovered, new interactions

AMO measurements provide strong limits on electron EDMs, parity-violating couplings, variation of constants, Lorentz violation, ultralight fields, and spin-dependent or composition-dependent forces. No result reviewed here is accepted as a confirmed discovery of a new force or dark-matter field.

A common static parameterization is a Yukawa correction,

$$V(r) = -\frac{Gm_1m_2}{r} \left[ 1+\alpha_Y e^{-r/\lambda} \right],$$

where $\alpha_Y$ is a model-dependent strength relative to gravity and $\lambda$ is the interaction range. A boson of mass $m_b$ has a Compton range

$$\lambda = \frac{\hbar}{m_b c}.$$

Spin-dependent interactions require different operators and source polarizations; they cannot be summarized by one scalar $\alpha_Y$.

Ultralight dark-matter searches instead use oscillatory or stochastic field models. Clocks may respond through effective constants, magnetometers through spin couplings, and atom interferometers through masses, transition frequencies, or differential acceleration. Limits are conditional on the field’s local density, coherence, polarization, coupling basis, and signal statistics.

What Is Actively Developing

International optical-clock networks

The immediate clock frontier combines:

• independent frequency-ratio measurements across laboratories;
• optical-fibre links over continental distances;
• improved intercontinental comparison;
• reliable contributions of optical standards to TAI;
• transportable clocks for link validation and geopotential surveys;
• robust clock-laser transfer and synchronous interrogation; and
• governance for a redefined second that remains accessible.

Current GNSS time transfer is not sufficient to expose the full performance of the best optical clocks at practical averaging times. Fibre links, transportable clocks, advanced two-way methods, and space links are active solutions with different geographic and infrastructure limits.

Quantum enhancement with complete overhead

Spin squeezing, entangled ensembles, collective cavity measurements, and adaptive protocols can increase Fisher information. The research question is whether they improve a complete sensor under matched conditions.

For two protocols, a fair comparison uses

$$\mathcal G_{\mathrm{wall}} = \frac{ \dot F_{\mathrm{quantum}} }{ \dot F_{\mathrm{reference}} } = \frac{ F_{\mathrm{quantum}}/T_{c,\mathrm{quantum}} }{ F_{\mathrm{reference}}/T_{c,\mathrm{reference}} }.$$

It must also compare dynamic range, loss, calibration, robustness, and systematic susceptibility. A sub-standard-quantum-limit state is an established resource when verified. End-to-end advantage is a separate claim.

Long-baseline and distributed atom interferometry

Long baselines increase interrogation time and enable gradiometric rejection of laser phase noise in proposed searches for mid-band gravitational waves and ultralight fields. Active obstacles include:

• high-flux, low-temperature atom sources;
• efficient single-photon beam splitters on clock transitions;
• large momentum transfer with controlled diffraction phases;
• wavefront characterization over long propagation distances;
• differential scale-factor and timing matching;
• gravity-gradient and Coriolis control;
• projection-noise reduction at large atom number;
• vibration and platform motion;
• vacuum, laser power, and alignment over the baseline; and
• a validated global likelihood for broadband and line-like searches.

Prototype success should update priors about feasibility, not erase the remaining engineering and inference chain.

Fieldable quantum sensors

Deployable gravimeters, magnetometers, clocks, and Rydberg electric-field sensors are moving from controlled laboratories into mobile, industrial, biomedical, geophysical, and space settings. Field performance adds variables that are easy to suppress indoors:

$$\mathbf z(t) = \left( \mathbf a,\, \boldsymbol\Omega,\, \mathbf B,\, T,\, p,\, \text{vibration},\, \text{orientation},\, \text{electromagnetic interference} \right).$$

A field-ready claim needs a specified envelope in $\mathbf z$, not one outdoor demonstration. Hybridization with classical sensors is often an advantage: a classical accelerometer can bridge phase wraps while an atom interferometer controls long-term bias.

Multiparameter and network inference

Many sensors respond to several parameters. For measurements $\mathbf y=\mathbf K\boldsymbol\theta+\boldsymbol\epsilon$, identifiability depends on the rank and conditioning of $\mathbf K$. A large sensitivity to one combination does not identify all components.

Networks add spatial and temporal covariance. They can reject local disturbances, triangulate a propagating transient, or seek a coherent field, but shared oscillators, transfer links, magnetic environments, and analysis pipelines can create common-mode noise. A network must publish or model its cross-spectral covariance, not only individual sensor floors.

Better theory as a sensing resource

Atomic, molecular, nuclear, and hadronic calculations translate measured shifts into constants or effective couplings. Active needs include:

• uncertainty estimates for relativistic many-body calculations;
• correlation among sensitivity coefficients;
• nuclear-size and nuclear-polarization corrections;
• response matrices with several low-energy operators;
• blind comparisons among independent computational methods; and
• data formats that permit global reinterpretation when theory improves.

A lower experimental noise floor can expose theory uncertainty rather than new physics. That is still progress if the discrepancy is reported at the correct layer.

What Is Debated or Uncertain

What counts as quantum advantage?

All AMO sensors rely on quantum energy levels or interference in a broad sense. A stronger claim of quantum-enhanced sensing normally means performance beyond a specified classical or unentangled resource bound. There is no useful platform-independent advantage number unless particle number, interrogation time, energy, bandwidth, prior information, and overhead are fixed.

How should platform readiness be ranked?

Laboratory sensitivity, calibration accuracy, size, mass, power, bandwidth, startup time, maintenance, environmental robustness, and cost are incommensurate objectives. A Pareto frontier is more honest than a total ranking. “Best quantum sensor” is usually not a scientific statement.

Will long-baseline atom interferometers reach discovery sensitivity?

The physical response models are credible and several ingredients have been demonstrated. Full-scale sensitivity depends on simultaneous control of atom flux, laser noise, wavefronts, baseline geometry, backgrounds, and long-term operation. Projected exclusion curves are conjectural instrument forecasts, not present limits.

How much of a new-force limit is experimental?

A detector constrains an observable. Converting it to a boson mass and coupling can require assumptions about source composition, shielding, screening, local dark-matter density, field coherence, astrophysical history, and operator dominance. Different assumptions can produce different limits from the same data.

When does a discrepancy become an anomaly?

A residual is scientifically interesting when:

1. its covariance and global significance are established;
2. the analysis was not selected after inspecting the feature;
3. known systematic and theory alternatives are quantitatively tested;
4. the feature predicts new observables; and
5. an independent apparatus can reproduce those observables.

The word “anomaly” should not skip these steps.

Key Experimental Platforms

Platform	Strength	Dominant frontier bottleneck	Representative target
optical lattice clock	many atoms, high stability, species comparisons	collisions, blackbody and lattice shifts, clock laser, transfer	time, geodesy, constants, ultralight fields
trapped-ion clock	excellent isolation and systematic control	single-ion stability, micromotion, logic operations	time, ratios, relativistic and new-physics tests
vapour-cell magnetometer	high sensitivity, arrays, room-temperature operation	shielding, heading error, gradients, bandwidth	biomagnetism, NMR, spin couplings
cold-atom interferometer	calculable matter-wave scale factor and inertial response	vibration, wavefronts, atom flux, dynamic range	gravity, rotation, equivalence, long-baseline searches
molecular beam or trap	large internal effective fields and diverse sensitivities	flux, state control, coherence, theory response	EDMs, parity violation, mass-ratio and dark-field searches
Rydberg atom sensor	large electric dipoles and broad RF/mm-wave response	calibration, dynamic range, spatial averaging, packaging	electromagnetic-field metrology
distributed clock or sensor network	common signals and spatial discrimination	links, synchronization, covariance, uptime	time scales, transients, geodesy, coherent fields

The table describes complementary operating points. It is not a maturity ranking.

Key Theoretical and Statistical Tools

Sensitivity functions and transfer functions

Time-domain sensitivity functions expose pulse timing, dead time, and aliasing. Frequency-domain transfer functions expose accepted signal bands and environmental feedthrough.

Fisher information and Bayesian design

Fisher information supports local precision bounds and experimental design. Bayesian methods can incorporate prior ranges, latent calibration parameters, and hierarchical network models. Neither framework removes the need to test model misspecification.

Allan and cross-spectral statistics

Allan-type variances diagnose clock and sensor stability under different noise colours. Cross spectra and coherence distinguish local noise from a shared signal in networks. Irregular sampling and data gaps modify the spectral window.

Reversal channels

Electric-field, magnetic-field, momentum, isotope, orientation, and interferometer reversals sort contributions by parity. A reversal projects onto a signature; it does not guarantee that only the target occupies that signature.

Effective field theory

Low-energy operators connect AMO observables to new interactions. A global fit can be written

$$\mathbf y = \mathbf A\mathbf c + \boldsymbol\epsilon,$$

where $\mathbf c$ contains effective couplings and $\mathbf A$ combines atomic, molecular, nuclear, and experimental response coefficients. Setting all but one coefficient to zero gives a one-source limit, not a universal constraint.

Blind and simulation-based validation

Hidden offsets, frozen cuts, signal injections, coverage tests, null streams, and synthetic data challenge the full inference chain. Open data and code are especially valuable for searches whose signal model may be reinterpreted later.

Evidence Required for a Frontier Claim

A precision record

Report the measurand, correction model, covariance, coverage convention, stability, uptime, averaging interval, and date. Compare like with like: systematic uncertainty is not instability.

A quantum-enhancement claim

Report the resource count, reference protocol, state-verification method, contrast, losses, cycle times, calibration overhead, and task-level performance on independent data.

A field-ready claim

Report the operating envelope, motion, orientation, temperature, vibration, electromagnetic environment, restart frequency, operator intervention, and comparison with an established instrument.

A fundamental-physics limit

Report the raw observable or sufficient statistics, likelihood, nuisance parameters, trials correction, signal injections, environmental vetoes, theory response, coupling assumptions, and confidence or credible-interval construction.

A discovery claim

Require independent reproduction and model-specific predictions beyond the data used to define the feature. A local excess from one instrument is not enough.

Common Misconceptions

The smallest noise floor is the most accurate sensor

A noise floor describes fluctuations. Accuracy concerns bias relative to a reference or measurement model. A quiet sensor can be wrong.

More interrogation time always improves sensitivity

Longer interrogation can increase phase while reducing contrast, bandwidth, dynamic range, and duty cycle. It can also increase exposure to drift and gradients.

Entanglement removes the standard quantum limit for free

Entanglement can change a statistical bound under declared resources. It does not remove loss, decoherence, dead time, readout error, or systematic uncertainty.

A clock at $10^{-19}$ resolves height at the millimetre level instantly

Near Earth’s surface, $\Delta\nu/\nu\simeq g\Delta h/c^2$, so a millimetre corresponds to about $10^{-19}$. Reaching that statistical and systematic comparison uncertainty, knowing the clock coordinates, transferring the frequency, and modelling tides and geopotential are separate requirements.

A long baseline automatically cancels laser noise

Differential interrogation can reject common phase noise when timing, propagation, pulse areas, contrasts, and scale factors match. Residual wavefront and propagation effects grow in importance with baseline.

A five-sigma discrepancy is automatically new physics

The quoted significance is conditional on a model and trials set. Unrecognized systematics, underestimated covariance, and theory error can also produce a discrepancy. Independent predictive confirmation is needed.

Fundamental constants are whatever the latest paper reports

CODATA values come from a correlated adjustment of compatible input data. New measurements enter later adjustments with their correlations and consistency tests.

Connections to Other Volumes

• Fisher Information supplies the score, Cramér–Rao inequality, and multiparameter Fisher matrix.
• Density Operators supplies the state language for noisy and partially polarized probes.
• Quantum Channels and Noise supplies loss, noise, and measurement-map language.
• Continuous Quantum Measurement supplies filtering, records, and backaction for monitored sensors.
• Time Reversal separates antiunitary symmetry from laboratory switch reversals.
• Fundamental Symmetry Frontiers tracks the dated evidence for EDMs, parity violation, nuclear moments, isotope-shift anomalies, and spin-dependent forces.
• Approximation and Error Estimates supplies controlled-model and validity language.
• Computational Reproducibility Benchmarks supplies versioning, convergence, and benchmark discipline.
• Frontiers and Open Problems supplies the chapter-wide claim taxonomy and archive policy.

What Changed This Year

The 2026 review adds four concrete updates.

1. Optical-clock ratios: a July 2026 Physical Review Letters article reported Al$^{+}$/Yb, Al$^{+}$/Sr, and Yb/Sr ratios at or below $3.2\times10^{-18}$ total fractional uncertainty. Its noted differences from an earlier campaign reinforce, rather than weaken, the need for repeated independent ratios and closure.
2. Long-baseline atom-interferometer ingredients: the June 2026 AION prototype demonstrated differential clock-transition interferometry consistent with its projection-noise model under large synthetic laser phase noise. Full long-baseline discovery sensitivity remains active.
3. Metrological governance: the BIPM process remains directed toward a possible optical redefinition of the second no earlier than 2030, with transfer, reliability, comparisons, and access treated as mandatory system properties.
4. Constants cycle: CODATA 2022 remains the current recommendation while the 2026 adjustment cycle is under way. New measurements should be cited as inputs or independent results until the adjustment is released.

This review does not identify a confirmed AMO discovery of a new force, variation of a constant, dark-matter field, or symmetry violation beyond the established Standard Model effects.

Representative Papers and Reviews

General quantum sensing

• C. L. Degen, F. Reinhard, and P. Cappellaro, “Quantum sensing,” Reviews of Modern Physics 89, 035002 (2017). A platform-spanning account of quantum-sensor protocols, sensitivity, and noise.
• L. Pezzè, A. Smerzi, M. K. Oberthaler, R. Schmied, and P. Treutlein, “Quantum metrology with nonclassical states of atomic ensembles,” Reviews of Modern Physics 90, 035005 (2018). The standard entry point for entanglement-assisted atomic metrology.

Clocks and standards

• A. D. Ludlow, M. M. Boyd, J. Ye, E. Peik, and P. O. Schmidt, “Optical atomic clocks,” Reviews of Modern Physics 87, 637–701 (2015).
• N. Dimarcq et al., “Roadmap towards the redefinition of the second,” Metrologia 61, 012001 (2024), together with the current BIPM redefinition FAQ.
• M. C. Marshall et al., “High-stability single-ion clock with $5.5\times10^{-19}$ systematic uncertainty,” Physical Review Letters 135, 033201 (2025).
• A. Aeppli et al., “Atomic clock frequency ratios with fractional uncertainty $\leq3.2\times10^{-18}$,” Physical Review Letters 137, 033201 (2026).

Atom interferometry and magnetometry

• A. D. Cronin, J. Schmiedmayer, and D. E. Pritchard, “Optics and interferometry with atoms and molecules,” Reviews of Modern Physics 81, 1051–1129 (2009).
• K. Bongs et al., “Taking atom interferometric quantum sensors from the laboratory to real-world applications,” Nature Reviews Physics 1, 731–739 (2019).
• K. Stolzenberg et al., “Multi-axis inertial sensing with 2D matter-wave arrays,” Physical Review Letters 134, 143601 (2025).
• C. F. A. Baynham et al., “A prototype differential atom interferometer for fundamental physics,” Nature 654, 622–628 (2026). The article also provides public data and synthetic-data validation.
• D. Budker and M. Romalis, “Optical magnetometry,” Nature Physics 3, 227–234 (2007).

Constants and new interactions

• E. Tiesinga, P. J. Mohr, D. B. Newell, and B. N. Taylor, “CODATA recommended values of the fundamental physical constants: 2022,” Journal of Physical and Chemical Reference Data 54, 033105 (2025), and the versioned NIST constants database.
• L. Morel, Z. Yao, P. Cladé, and S. Guellati-Khélifa, “Determination of the fine-structure constant with an accuracy of 81 parts per trillion,” Nature 588, 61–65 (2020).
• M. S. Safronova et al., “Search for new physics with atoms and molecules,” Reviews of Modern Physics 90, 025008 (2018).
• D. DeMille, N. R. Hutzler, A. M. Rey, and T. Zelevinsky, “Quantum sensing and metrology for fundamental physics with molecules,” Nature Physics 20, 741–749 (2024).
• S. D. Bass and M. Doser, “Quantum sensing for particle physics,” Nature Reviews Physics 6, 329–339 (2024).

Exercises

Exercise 1: Information rate, not single-shot gain

Protocol A uses uncorrelated atoms, has Fisher information $F_A=1.0\times10^4$ per shot, and cycles every $1.0\ \mathrm{s}$. Protocol B uses a squeezed state, has $F_B=2.5\times10^4$ per accepted shot, takes $2.0\ \mathrm{s}$ per attempted cycle, and accepts $80\%$ of attempts.

1. Compute the Fisher-information rate for each protocol.
2. Which has the smaller ideal statistical variance after one hour?
3. By what factor?

Solution

For A,

$$\dot F_A = \frac{10^4}{1.0\ \mathrm{s}} = 1.0\times10^4\ \mathrm{s}^{-1}.$$

For B, the accepted information per attempt is $0.80F_B=2.0\times10^4$, so

$$\dot F_B = \frac{2.0\times10^4}{2.0\ \mathrm{s}} = 1.0\times10^4\ \mathrm{s}^{-1}.$$

The protocols have equal ideal information rate and therefore equal Cramér–Rao variance after one hour. Protocol B has a single-shot resource gain but no wall-clock statistical gain under the stated overhead. Their systematic errors and dynamic ranges would still need comparison.

Exercise 2: Atom-interferometer scale and phase ambiguity

An ideal light-pulse accelerometer has $k_{\mathrm{eff}}=1.61\times10^7\ \mathrm{m}^{-1}$ and $T=0.20\ \mathrm{s}$.

1. Find the phase change caused by $\delta a=1.0\times10^{-8}\ \mathrm{m\,s^{-2}}$.
2. Find the acceleration interval corresponding to a $2\pi$ phase change.
3. Explain why sensitivity and dynamic range must be reported together.

Solution

Using $\delta\phi=k_{\mathrm{eff}}\delta aT^2$,

$$\delta\phi = (1.61\times10^7) (1.0\times10^{-8}) (0.20)^2 = 6.44\times10^{-3}\ \mathrm{rad}.$$

One $2\pi$ period corresponds to

$$\Delta a_{2\pi} = \frac{2\pi} {k_{\mathrm{eff}}T^2} \approx 9.76\times10^{-6}\ \mathrm{m\,s^{-2}}.$$

Increasing $T$ amplifies a small acceleration but shrinks the unambiguous acceleration interval as $T^{-2}$. A high-sensitivity interferometer may need a chirp, midfringe lock, auxiliary accelerometer, multiple scale factors, or phase-unwrapping model in a dynamic environment.

Exercise 3: Correlated clock-network uncertainty

Two clock ratios $r_1$ and $r_2$ each have standard uncertainty $u=3.0\times10^{-18}$ and correlation coefficient $\rho=0.60$ because they share an oscillator and link. Find the standard uncertainty of:

1. the average $(r_1+r_2)/2$;
2. the difference $r_1-r_2$.

Treat the ratios as small fractional deviations around nominal values.

Solution

The covariance is $\operatorname{Cov}(r_1,r_2)=\rho u^2$. Therefore

$$\operatorname{Var} \left( \frac{r_1+r_2}{2} \right) = \frac{u^2+u^2+2\rho u^2}{4} = \frac{u^2}{2}(1+\rho).$$

Thus

$$u_{\mathrm{avg}} = 3.0\times10^{-18} \sqrt{0.8} \approx 2.68\times10^{-18}.$$

For the difference,

$$\operatorname{Var}(r_1-r_2) = 2u^2(1-\rho),$$

so

$$u_{\mathrm{diff}} = 3.0\times10^{-18} \sqrt{0.8} \approx 2.68\times10^{-18}.$$

The equal numerical values are specific to $\rho=0.60$. Positive common-mode correlation weakens averaging but improves a difference relative to the uncorrelated case.

Exercise 4: Range of an ultralight mediator

Find the Compton interaction range for a boson with mass $m_b=10^{-12}\ \mathrm{eV}/c^2$. Use $\hbar c=1.97327\times10^{-7}\ \mathrm{eV\,m}$. Name two reasons why this range alone does not determine an experiment’s sensitivity.

Solution

Because

$$\lambda = \frac{\hbar c}{m_bc^2},$$

the range is

$$\lambda = \frac{1.97327\times10^{-7}\ \mathrm{eV\,m}} {10^{-12}\ \mathrm{eV}} = 1.97327\times10^5\ \mathrm{m} \approx 197\ \mathrm{km}.$$

Sensitivity also depends on the coupling strength and operator, source mass or spin density, geometry, composition, screening, background gradients, sensor transfer function, and integration time. Equal Compton range does not make scalar, vector, and spin-dependent interactions experimentally equivalent.

Exercise 5: Clock-ratio response to a varying constant

A clock ratio has differential sensitivity $K_\alpha^{(A)}-K_\alpha^{(B)}=6.0$. A search limits a coherent fractional ratio oscillation to amplitude $1.2\times10^{-18}$ at a selected frequency. Assuming only $\alpha$ varies and the transfer function is unity, find the corresponding amplitude limit on $\delta\alpha/\alpha$. List three assumptions hidden by the phrase “assuming only $\alpha$ varies.”

Solution

The linear response is

$$\frac{\delta R}{R} = \left( K_\alpha^{(A)}-K_\alpha^{(B)} \right) \frac{\delta\alpha}{\alpha}.$$

Therefore

$$\left| \frac{\delta\alpha}{\alpha} \right| < \frac{1.2\times10^{-18}}{6.0} = 2.0\times10^{-19}.$$

Hidden assumptions include that other effective constants do not vary or their couplings vanish; the calculated sensitivity coefficients and their correlations are adequate; the field is coherent over the measurement; the sampling transfer function is correctly modelled; the local field density or normalization is known; and environmental oscillations at that frequency have been excluded.

Exercise 6: Noise spectral density and integration

A magnetometer has white magnetic-noise amplitude $S_B^{1/2}=15\ \mathrm{fT}/\sqrt{\mathrm{Hz}}$ over the relevant band. For a known-phase sinusoidal signal and an ideal matched estimator, estimate the statistical amplitude uncertainty after $\tau=400\ \mathrm{s}$ using $u_B\simeq S_B^{1/2}/\sqrt{\tau}$. Why is this not automatically a detection limit for an unknown-frequency signal?

Solution

The ideal uncertainty is

$$u_B \simeq \frac{15\ \mathrm{fT}/\sqrt{\mathrm{Hz}}} {\sqrt{400\ \mathrm{s}}} = 0.75\ \mathrm{fT}.$$

An unknown-frequency search scans multiple frequencies, phases, and possibly time windows. It must account for the spectral window, nonwhite noise, calibration, line broadening, trials factor, and detection efficiency. Environmental lines can also occupy the same band. The one-template uncertainty is therefore not a global discovery threshold.

Exercise 7: Interpret a prototype milestone

A differential atom-interferometer prototype reaches its modelled projection-noise limit while several radians of synthetic common laser phase noise are applied. State one established conclusion, three active questions, and one conclusion that would be an overclaim.

Solution

An established conclusion is that the demonstrated differential configuration and estimator reject the applied common phase noise to the reported measurement resolution under the prototype’s atom number, contrasts, geometry, timing, and signal model.

Active questions include whether the rejection persists at longer baselines where propagation and wavefront errors matter; whether higher atom number and lower projection noise expose new residuals; whether large momentum transfer, long-term alignment, backgrounds, and duty cycle meet a full detector budget; and whether the broadband search likelihood has validated coverage under real nonstationary noise.

It would be an overclaim to say that the prototype detected a gravitational wave or dark matter, or that a full-scale observatory is already guaranteed to reach its projected sensitivity.