Generalized Clock-Noise Observables

• Generalized clock-noise observables are defined operationally, encoding timing mismatches and noise via relational protocols across various physical frameworks.
• They are constructed using gauge-invariant protocols and statistical estimators that separate deterministic clock signals from stochastic or instrumental noise.
• Their applications span calibrating cosmic clocks, refining quantum timing in interferometry, and enabling clock-noise subtraction in space-based gravitational-wave metrology.

Generalized clock-noise observables are operational quantities that encode timing, proper-time mismatch, clock-induced coherence loss, oscillator fluctuations, or calibration residuals in systems where clocks are part of the measurement architecture. In the literature considered here, they appear in several technically distinct forms: the cosmic-clock field $T(\tilde x)$ comparing constant proper-time and constant observed-redshift slicings in relativistic large-scale structure; the diffeomorphism-invariant time delay $\delta\tau(s)=s-\tau(s)$ in linearized quantum gravity; the path visibility $\mathcal V$ of noisy quantum clocks in interferometry; the generalized link observables $r_{\pm i,\pm j}$ used for clock-noise subtraction in geometric time-delay interferometry; quadratic variance estimators for power-law clock noise; and the paired Ramsey error signals used in generalized auto-balanced Ramsey spectroscopy (Jeong et al., 2014, Bonga et al., 2013, Verma et al., 2020, Luo et al., 10 Jul 2026, Ashby, 2011, Yudin et al., 2017).

Domain	Observable	Role
Relativistic cosmology	$T(\tilde x)$	Proper-time perturbation at fixed observed redshift
Linearized quantum gravity	$\delta\tau(s)$	Diffeomorphism-invariant timing observable
Quantum-clock interferometry	$\mathcal V=|\kappa|$	Off-diagonal path coherence under noisy clock evolution
Space-based GW interferometry	$r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$	Clock-noise subtraction in geometric TDI
Oscillator metrology	$A=U^T H U$	Allan-type quadratic estimator of clock noise
Ramsey spectroscopy	$S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$	Rejection of probe-induced clock perturbations

1. Operational definition and recurring structure

Across these settings, the defining feature of a generalized clock-noise observable is not a common physical platform but a common construction principle: a clock-related quantity is defined relationally, then embedded in a perturbative, statistical, or control-theoretic framework that distinguishes physical signal from contamination or bias. In cosmology, the relevant comparison is between a proper-time slicing and an observed-redshift slicing. In quantum gravity, it is between two timelike proper times constrained by a null signal. In interferometry, it is the coherence between arm-conditioned clock states. In TDI, it is the measurable combination of delayed or advanced onboard clock jitters. In frequency metrology, it is a quadratic form built from filtered time-error data. In Ramsey spectroscopy, it is a discriminator engineered so that interrogation-induced perturbations are absorbed by an auxiliary control variable rather than mapped into the clock frequency channel (Jeong et al., 2014, Bonga et al., 2013, Verma et al., 2020, Luo et al., 10 Jul 2026, Ashby, 2011, Yudin et al., 2017).

Taken together, these works suggest a common architecture. First, the observable is specified by a physical protocol: signal exchange, light propagation, interferometric recombination, sideband comparison, differencing of oscillator records, or phase-jump Ramsey interrogation. Second, the observable is written in a form that is invariant under irrelevant description choices, such as gauge in relativistic perturbation theory or coordinate diffeomorphisms in the timing thought experiment. Third, noise is either computed explicitly from a stochastic or quantum model, represented statistically through quadratic forms and eigenvalues, or rejected by design through subtraction or servo balancing. A persistent theme is that the clock-related quantity itself is often not the noise: the deterministic relativistic or dynamical clock signal must be defined first, and only then can environmental, instrumental, or estimator noise be meaningfully added or removed.

2. Relativistic cosmic clocks and proper-time mismatch

In relativistic large-scale structure, the central clock observable is the cosmic-clock field

$\delta\tau(s)=s-\tau(s)$0

where $\delta\tau(s)=s-\tau(s)$1 is the source proper time, $\delta\tau(s)=s-\tau(s)$2 is the corresponding background conformal time, and $\delta\tau(s)=s-\tau(s)$3 is the scale factor inferred from observed redshift. The observable is explicitly the difference in $\delta\tau(s)=s-\tau(s)$4 between a constant-proper-time hypersurface and a constant-observed-redshift hypersurface. At linear order it is also a proper-time perturbation,

$\delta\tau(s)=s-\tau(s)$5

so it measures the local age perturbation at fixed observed redshift (Jeong et al., 2014).

The formalism is developed in a linearly perturbed flat FRW spacetime with scalar, vector, and tensor perturbations,

$\delta\tau(s)=s-\tau(s)$6

The observed redshift perturbation enters through

$\delta\tau(s)=s-\tau(s)$7

with

$\delta\tau(s)=s-\tau(s)$8

The cosmic-clock observable then becomes

$\delta\tau(s)=s-\tau(s)$9

Its two terms have distinct meanings: $\mathcal V$0 is the mismatch induced by gravitational and Doppler redshift perturbations plus line-of-sight propagation effects, while the proper-time integral converts coordinate-time displacement into a perturbation of the source clock.

The formalism identifies two classes of cosmic clocks: sharp events at a unique time, such as recombination, neutrino decoupling, BBN, and baryon thermal decoupling on sufficiently large scales; and objects or rulers with known time evolution, such as evolving galaxy sizes or evolving tracer correlation scales. On sufficiently large scales, the CMB Sachs–Wolfe limit appears as $\mathcal V$1. More generally, $\mathcal V$2 feeds directly into relativistic ruler distortions and observed number counts. For evolving rulers,

$\mathcal V$3

and for tracer density,

$\mathcal V$4

A central clarification is that $\mathcal V$5 is a physical relativistic clock perturbation, not a noise term. A plausible extension is to append stochastic or instrumental contamination only after this gauge-invariant signal model has been defined.

3. Diffeomorphism-invariant time delay and quantum timing noise

A second major construction arises from a clock synchronization and signal-exchange thought experiment in a lab-equipped spacetime $\mathcal V$6. A lab follows a timelike geodesic, ejects a freely falling probe at event $\mathcal V$7, synchronizes both clocks there, and later receives a timestamped electromagnetic signal from the probe. If the signal is received at lab proper time $\mathcal V$8 and was emitted when the probe clock read $\mathcal V$9, the observable is

$r_{\pm i,\pm j}$0

Because it is defined by relational data and protocol rather than coordinates, it is invariant under diffeomorphisms acting simultaneously on all elements of the lab-equipped spacetime data (Bonga et al., 2013).

In Minkowski background, the classical configuration is a geodesic triangle $r_{\pm i,\pm j}$1 composed of lab, probe, and null-signal segments. The classical emission time is

$r_{\pm i,\pm j}$2

where the rapidity $r_{\pm i,\pm j}$3 is fixed by the lab and probe 4-velocities. In linearized gravity the observable is decomposed as

$r_{\pm i,\pm j}$4

with $r_{\pm i,\pm j}$5 a linear functional of the graviton field expressed as a sum of segment integrals and iterated affine integrals. This reduction is the key step that makes the observable calculable and extensible to related timing, ranging, and astrometric quantities.

Upon quantization in the Minkowski Fock vacuum of linearized gravity, the mean linear correction vanishes, $r_{\pm i,\pm j}$6, while the leading nontrivial fluctuation comes from $r_{\pm i,\pm j}$7. The unsmeared variance is UV divergent because the Hadamard two-point function behaves like $r_{\pm i,\pm j}$8. The paper therefore introduces detector-resolution smearing with characteristic width $r_{\pm i,\pm j}$9, replacing the point field by a smeared field. The regularized variance then scales as

$T(\tilde x)$0

This is a prototype result rather than a definitive phenomenological prediction, because the calculation omits the quadratic correction $T(\tilde x)$1, uses a non-Lorentz-invariant smearing model, and exhibits a puzzling low-velocity divergence. The authors explicitly treat the divergence as provisional, possibly arising from the approximation regime, detector modeling, or neglected second-order terms.

4. Noisy quantum clocks and interferometric coherence

In quantum-clock interferometry, the observable of interest is not a time delay but path visibility. A massive particle with internal dynamical degrees of freedom acts as a clock because its internal state is prepared in a superposition that is not an eigenstate of $T(\tilde x)$2. Along a path with proper time $T(\tilde x)$3, the clock state is

$T(\tilde x)$4

In a Mach–Zehnder interferometer with arm proper times $T(\tilde x)$5 and $T(\tilde x)$6, clock evolution entangles internal state and path, reducing interference visibility in accordance with $T(\tilde x)$7 (Verma et al., 2020).

The generalized noisy-clock observable is obtained by coupling the clock to an environment and tracing out clock and environment after arm-dependent evolution. If the joint clock–environment states on the two arms are $T(\tilde x)$8 and $T(\tilde x)$9, the reduced path density matrix contains off-diagonal coherence proportional to

$\delta\tau(s)$0

and the detector probabilities are

$\delta\tau(s)$1

Hence the generalized visibility is

$\delta\tau(s)$2

This extends the noiseless formula $\delta\tau(s)$3. A crucial point is that, in the noisy case, visibility is not generally a simple overlap of reduced clock states such as $\delta\tau(s)$4; it is controlled by the arm coherence inherited from the joint clock–environment evolution.

The behavior of this observable is regime dependent. For small noise and small proper-time difference, the visibility is further reduced by noise. In more general regimes, noise can either decrease or increase visibility. The paper identifies two competing effects: environmental coupling can corrupt the clock’s proper-time record and thereby reduce which-path information, but it can also enlarge the effective Hilbert space and encode additional arm information into the environment. Explicit models confirm both possibilities. A thermal Jaynes–Cummings environment lowers visibility as temperature increases, while standard quantum-channel models based on amplitude damping, phase damping, and depolarizing dynamics produce model-dependent nonmonotonic behavior. In the low-noise regime the ordering

$\delta\tau(s)$5

is reported, reflecting the increasing minimal environment dimension.

5. Generalized clock-noise observables in geometric TDI

In space-based millihertz gravitational-wave interferometry, laser phase noise is removed by TDI, but clock jitter from onboard ultra-stable oscillators can remain above the secondary-noise floor and bias likelihood weighting. The standard sideband-derived clock observable is

$\delta\tau(s)$6

with $\delta\tau(s)$7 the delay operator and $\delta\tau(s)$8 the clock noise on spacecraft $\delta\tau(s)$9. The key advance of geometric TDI is that once time-advance operators $\mathcal V=|\kappa|$0 are allowed, an arbitrary two-path observable requires four exhaustive local space-time link structures rather than the forward-time case alone (Luo et al., 10 Jul 2026).

Link structure	Observable	Result
Forward delay	$\mathcal V=|\kappa|$1	Standard case
Delay after time advance	$\mathcal V=|\kappa|$2	$\mathcal V=|\kappa|$3
Backward propagation to advanced local clock	$\mathcal V=|\kappa|$4	$\mathcal V=|\kappa|$5
Both clocks time-advanced	$\mathcal V=|\kappa|$6	$\mathcal V=|\kappa|$7

These generalized observables make the clock-noise residual algebraically parallel to the laser-noise residual. The clock part of each generalized $\mathcal V=|\kappa|$8 can be written as

$\mathcal V=|\kappa|$9

and the residual clock noise in an arbitrary two-path geometric TDI observable becomes a telescoping path sum built from delayed or advanced products of $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$0. The resulting clock-calibrated observable is

$r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$1

Conceptually, $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$2 plays for clock noise the same algebraic role that $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$3 plays for laser noise.

The framework is constructive rather than purely formal. For any two-path geometric TDI observable, one classifies each link by its local delay/advance structure, accumulates the corresponding clock term recursively, sums along both paths, and adds the explicitly measurable $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$4 terms. The paper works this out for first-generation Monitor-E, modified second-generation Michelson $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$5, and especially modified second-generation U-type $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$6, whose geometry genuinely requires the advanced-link observables. Time-domain simulations using LISA-like orbits and noise levels show that for $r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$7 the subtraction suppresses the clock-noise residual below the signal region, restores the expected sensitivity to a monochromatic source, and improves Fisher and MCMC constraints on source amplitude, frequency, and phase. In this setting, generalized clock-noise observables are not merely diagnostics; they are explicit calibration terms required for precision inference.

6. Statistical estimators and rejection observables in metrology

In oscillator metrology, generalized clock-noise observables take the form of variance estimators built from filtered time-error records. For a discrete clock process with one-sided fractional-frequency spectrum

$r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$8

Ashby constructs simulations by shaping white Fourier coefficients with exponent

$r_{i,j},r_{i,-j},r_{-i,j},r_{-i,-j}$9

The simulated time residual sequence $A=U^T H U$0 is then processed by linear filters such as the normalized second difference

$A=U^T H U$1

For the overlapping Allan variance, the expected value obeys

$A=U^T H U$2

and the finite-record estimator can be written as a quadratic form

$A=U^T H U$3

Its mean is the trace,

$A=U^T H U$4

its variance is

$A=U^T H U$5

and its full probability law is a generalized chi-square distribution determined by the eigenvalues $A=U^T H U$6. This matrix-and-eigenvalue formulation extends beyond Allan variance to modified Allan, Hadamard, Theo, and dead-time observables, making it a general statistical template for clock-noise estimation (Ashby, 2011).

A different metrological generalization appears in generalized auto-balanced Ramsey spectroscopy. Here the disturbance is not stochastic oscillator noise but a measurement-induced shift of the clock transition during Ramsey pulses. GABRS introduces two interleaved Ramsey sequences with dark times $A=U^T H U$7 and $A=U^T H U$8, producing two error signals,

$A=U^T H U$9

where $S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$0 is the LO detuning and $S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$1 is a concomitant pulse parameter. The phase-jump discriminator has the factorized form

$S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$2

so relaxation enters only as a multiplicative attenuation. The coupled lock conditions

$S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$3

always admit the exact shift-free solution

$S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$4

Among several variants, the most important uses an additional frequency step during both Ramsey pulses as the concomitant variable. In that case the secondary loop yields

$S_{T_1}^{\rm(err)},S_{T_2}^{\rm(err)}$5

and the discriminator acquires a universal anti-symmetric form at finite modulation amplitude. Other variants, based on an added phase shift or variable second-pulse duration, retain the exact zero-shift lock point but do not have the same universal discriminator shape (Yudin et al., 2017).

These two metrological strands occupy opposite ends of the clock-noise problem. The quadratic-estimator framework analyzes stochastic fluctuations already present in a clock record, whereas GABRS engineers observables that prevent interrogation-induced perturbations from entering the clock-frequency channel at all. Their conjunction clarifies a broader principle: generalized clock-noise observables can serve either as statistical estimators of noise or as control observables that project bias into a separate, measurable auxiliary degree of freedom.