Line-of-Sight Acceleration in Astrophysics

• Line-of-sight acceleration is the projection of an object’s acceleration onto the observer’s view, crucial for probing gravitational potentials and environmental forces.
• It is measured via radial-velocity drifts in Galactic spectroscopy, pulsar timing, and gravitational-wave Doppler modulation, offering diverse observational avenues.
• Precision measurements and modeling across these channels help constrain Galactic structures, test gravity theories, and refine parameters of compact binaries.

Line-of-sight acceleration is the component of an object’s acceleration projected onto the observer’s line of sight. In Galactic dynamics it is the observable obtained from redshift drift or radial-velocity drift, in pulsar timing it is inferred from secular orbital-period derivatives, and in gravitational-wave astronomy it appears as a time-dependent Doppler modulation of the waveform phase and amplitude. Across these settings, the quantity is used to probe local potential gradients, compact-binary environments, and possible deviations from Newtonian gravity, with distinct measurement strategies but closely related kinematic definitions (Barbosa et al., 30 Jun 2025, Tiwari et al., 27 Jun 2025, Moran et al., 2023).

Context	Observable	Representative result
Galactic spectroscopy	$\Delta v = a_{\rm LOS}\Delta t$	165 globular clusters require $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$ for Yukawa tests; $1.3\times10^5$ RR Lyrae at $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$ match rotation-curve power
Pulsar timing	$a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$	29 binary pulsars yield $$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$
Gravitational waves	Doppler-induced phase and time remapping	Current catalog analyses find LOS acceleration consistent with zero

1. Definition, projection geometry, and basic observables

In its most direct form, line-of-sight acceleration is the projection of the full acceleration vector onto the observer’s line of sight. For a spherically symmetric gravitational potential $\Phi(r)$, the total acceleration is

$$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$

and the projected quantity is

$$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$

where $\hat{\mathbf u}$ is the line-of-sight unit vector and $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$0 is the angle between $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$1 and $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$2. In the idealized case of an observer at the Galactic center, $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$3 and $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$4; for a Solar-system observer, the projection factor must be retained (Barbosa et al., 30 Jun 2025).

The direct electromagnetic observable is radial-velocity drift over a time baseline $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$5,

$$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$6

with the equivalent redshift-drift form $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$7. For a decade-long baseline, $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$8, and typical Galactic accelerations $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$9, next-generation spectrographs aim at $1.3\times10^5$0 precision or better (Barbosa et al., 30 Jun 2025).

In gravitational-wave analyses, the same physics is usually parameterized by

$1.3\times10^5$1

because only the Doppler modulation $1.3\times10^5$2 is observable. A constant line-of-sight velocity is not itself identifiable as an environmental effect: the constant-velocity term $1.3\times10^5$3 is completely degenerate with a re-definition of the coalescence time and a uniform mass rescaling, so LOSA inference targets the acceleration term (Pompili et al., 26 Jun 2026).

2. Galactic gravity and redshift-drift measurements

In Galactic applications, LOS acceleration is treated as a direct probe of the Milky Way potential. A four-component model has been used in which the total potential is

$1.3\times10^5$4

with a spherical bulge, two Miyamoto–Nagai disks, and a dark-matter halo with NFW density plus Yukawa correction. The projected observable is written as

$1.3\times10^5$5

and the Yukawa-corrected point-mass potential takes the form

$1.3\times10^5$6

Its radial derivative contains the Newtonian $1.3\times10^5$7 term together with the Yukawa “Coulomb” and Yukawa “gradient” contributions (Barbosa et al., 30 Jun 2025).

Forecasts based on 165 Milky Way globular clusters and $1.3\times10^5$8 RR Lyrae stars show that target multiplicity is decisive. For globular clusters, distances span $1.3\times10^5$9–$$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$0, typical Newtonian LOS accelerations range from a few $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$1 at large $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$2 up to $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$3 for the innermost clusters, and the largest $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$4 exceeds $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$5 only for five clusters near $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$6. With this sample, $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$7 precision is not useful, and rotation-curve fidelity is only matched for $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$8 for Yukawa parameters and $$\sigma_a \simeq 10\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$9 for Newtonian halo parameters (Barbosa et al., 30 Jun 2025).

For RR Lyrae stars, the accessible regime is substantially denser. Their distances span $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$0–$a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$1, LOS accelerations extend up to $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$2 at $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$3, and about $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$4 stars show $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$5. At $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$6, the inferred constraints on $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$7 are already as strong as, or stronger than, those from rotation curves. This suggests that in direct Galactic acceleration mapping, sample size can compensate for modest per-object precision (Barbosa et al., 30 Jun 2025).

The observational advantages and limitations are sharply defined. Direct LOS-acceleration measurements provide a model-independent probing of the gravitational field with no assumption of dynamical or virial equilibrium, are sensitive to local potential gradients, and are particularly powerful in the inner Galaxy where rotation-curve tracers are scarce. The corresponding limitations are ultra-high spectrographic stability over decade-long baselines, careful removal of solar-reflex motion and perspective acceleration terms, scatter introduced by the projection factor $a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$8, and systematics from stellar binaries and intrinsic jitter (Barbosa et al., 30 Jun 2025).

3. Pulsar timing and local acceleration-field reconstruction

Binary pulsars provide a distinct direct measurement of line-of-sight acceleration through secular orbital-period evolution. The measured quantity is decomposed as

$a_{\rm los}=c\,\dot P_b^{\rm Gal}/P_b$9

where the Shklovskii term is

$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$0

the gravitational-wave damping term is computed from the Peters–Mathews formula, and the Galactic piece is translated into line-of-sight acceleration via

$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$1

This is a differential Galactic acceleration between pulsar and Earth along the line of sight (Moran et al., 2023).

The first data release of direct line-of-sight acceleration measurements for 29 binary pulsars was presented by Moran et al. The individual accelerations span $$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$2–$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$3, with typical fractional uncertainties of $$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$4–$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$5. To interpret these data, a local first-order expansion of the Galactic acceleration field in cylindrical galactocentric coordinates was used,

$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$6

and the vertical coefficient was related to the local disk density by

$$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$7

The resulting inference was $$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$8 (Moran et al., 2023).

The same analysis also found evidence for unmodeled noise of unknown origin. A nuisance model in which any pulsar may carry an extra acceleration $$\rho_d = 0.040^{+0.020}_{-0.020}\,M_\odot\,\mathrm{pc}^{-3}$$9 drawn from a power law yielded posteriors $\Phi(r)$0 and $\Phi(r)$1. This indicates that direct LOS-acceleration catalogs are already sensitive enough that third bodies, accretion, or other unmodeled processes must be incorporated in the statistical model rather than treated as negligible perturbations (Moran et al., 2023).

4. Gravitational-wave imprint: phase corrections, harmonics, and time-domain remapping

For compact binary coalescences, a center-of-mass LOS acceleration produces a time-varying Doppler shift in the observed waveform. In the stationary-phase approximation, the inspiral phase acquires an extra term

$\Phi(r)$2

where the non-spinning point-particle, aligned-spin, and tidal-deformability contributions all start at $\Phi(r)$3 relative to the quadrupole and extend through $\Phi(r)$4. A compact factorized expression is

$\Phi(r)$5

with $\Phi(r)$6 and $\Phi(r)$7. The leading term arises from the time-varying Doppler shift induced by constant acceleration and breaks the mass–redshift degeneracy (Tiwari et al., 27 Jun 2025).

A complementary derivation uses an exact time-domain map between source and observer time. Neglecting constant Rømer delays and assuming constant LOS acceleration,

$\Phi(r)$8

so that

$\Phi(r)$9

At leading order, $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$0, so the observed frequency is red- or blue-shifted according to the accumulated LOS velocity. This treatment captures both phase and amplitude Doppler effects and makes no small-$$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$1 or quasi-circular approximations (Pompili et al., 26 Jun 2026).

Higher harmonics and eccentricity are central to waveform fidelity. For quasicircular higher modes, each $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$2 harmonic has its own stationary time,

$$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$3

and therefore its own phase shift,

$$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$4

For eccentric binaries, each harmonic $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$5 acquires its own acceleration correction, with the leading eccentric phase shift scaling as $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$6 and carrying explicit eccentricity dependence. An inconsistent treatment of LOS acceleration between higher harmonics can lead to biased conclusions; the same is true if eccentricity is ignored while fitting for LOSA (Roy et al., 7 Jun 2026).

5. Inference pipelines, waveform systematics, and catalog constraints

Current inference strategies fall into two related classes. In frequency-domain implementations, LOSA corrections are inserted into the inspiral phase of models such as IMRPhenomXP_NRTidalv2 and IMRPhenomXP within the LVK collaboration’s flagship parameter-estimation software \textsc{Bilby_tgr}, with Dynesty nested sampling and a distance- and phase-marginalized relative-binning likelihood. A representative setup samples $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$7 together with standard sky, orientation, spin, or tidal parameters, using an added hyper-parameter $$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$8 with prior Uniform$$\mathbf{a}(r)=-\nabla\Phi(r)=-\frac{d\Phi}{dr}\,\hat{\mathbf r},$$9 (Tiwari et al., 27 Jun 2025).

Time-domain implementations instead apply the Doppler remap directly to the strain produced by waveform generators. This has been embedded in pySEOBNR after generation of $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$0 and $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$1, using SEOBNRv6EHM for aligned spins plus eccentricity and SEOBNRv5PHM for generic spin precession, quasi-circular evolution. The approach is model-agnostic, costs only one 1D interpolation per polarization per waveform, and preserves all mode content, precession effects, and eccentricity by construction. A parallel program has applied the same time-domain logic across the O1–O4 compact-binary catalog using PyCBC Inference with dynesty and waveform families selected by source class, including IMRPhenomXPHM, SEOBNRv5PHM, IMRPhenomNSBH, and tidal or eccentric variants where needed (Pompili et al., 26 Jun 2026, Roy et al., 24 Jun 2026).

Injection–recovery studies delimit the range of reliable LOSA inference. When injection and recovery waveform models are identical, LOSAs are recovered as expected. The principal LOSA mimickers are now well characterized: higher modes for asymmetric binaries with $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$2, strong precession with $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$3, eccentricity with $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$4, omitted tidal effects in BNS signals, and beyond-GR phase modifications such as dipole radiation or a massive graviton. These effects can shift the inferred $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$5 or $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$6 toward prior boundaries and bias masses and spins if the template omits the relevant physics (Tiwari et al., 27 Jun 2025).

The eccentricity–LOSA degeneracy has become a focal point. SEOBNRv6EHM recovers LOSA correctly on both eccentric and spin-precessing injections, while SEOBNRv5PHM yields a spurious LOSA measurement on eccentric signals. In real data, all five neutron-star–black-hole events analyzed with joint $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$7 inference in GWTC-4.0 O4b are consistent with $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$8, but for GW200105_162426 the joint posterior disfavors both $$a_{\rm LOS}(r)=\mathbf a(r)\cdot\hat{\mathbf u}=-\frac{d\Phi}{dr}\cos\alpha,$$9 and $\hat{\mathbf u}$0 being zero simultaneously at $\hat{\mathbf u}$1 credibility. This supports eccentricity hints while showing no strong evidence for LOSA once eccentricity is accounted for (Pompili et al., 26 Jun 2026).

Catalog-level observational constraints remain consistent with zero acceleration. The O1–O4 study finds the LOS acceleration for all known binaries to date is consistent with zero. Current ground-based observatories are sensitive enough to only constrain scenarios that produce high accelerations, with $\hat{\mathbf u}$2 for BBH sources and $\hat{\mathbf u}$3 for BNS sources. Specific measurements include GW170817, for which $\hat{\mathbf u}$4, and GW190425, for which $\hat{\mathbf u}$5; Bayesian evidence comparisons uniformly favor the zero-acceleration hypothesis (Roy et al., 24 Jun 2026). Earlier Bilby/dynesty analyses of the same BNS events reported $\hat{\mathbf u}$6 confidence intervals of $\hat{\mathbf u}$7 for GW170817 and $\hat{\mathbf u}$8 for GW190425 (Vijaykumar et al., 2023).

6. Astrophysical interpretation, misconceptions, and future reach

In compact-binary astrophysics, LOS acceleration is interpreted as an environmental diagnostic. A perturber of mass $\hat{\mathbf u}$9 at separation $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$00 produces

$$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$01

Current non-detections at $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$02 exclude only very close stellar or intermediate-mass perturbers and remain consistent with standard formation channels in globular clusters, where $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$03–$$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$04, or galactic nuclei, where $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$05–$$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$06. Exceptionally large $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$07 may arise in active-galactic-nucleus disk captures, for which the tail of the $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$08 distribution can reach $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$09 (Pompili et al., 26 Jun 2026).

A common misconception is that LOSA is primarily a measurement of line-of-sight velocity. In both Galactic and gravitational-wave settings, constant velocity is either absorbed into redefinitions of masses and times or treated as an unobservable offset; the measurable effect is the secular drift. A second misconception is that the LOSA phase correction is intrinsically unique. The present waveform literature shows that eccentricity, higher-order modes, precession, and omitted matter effects can mimic or mask a non-zero LOSA, so waveform accuracy is a leading concern for inference (Tiwari et al., 27 Jun 2025, Roy et al., 7 Jun 2026).

Near-term prospects diverge by observational channel. In Galactic spectroscopy, next-generation high-resolution spectrographs such as ANDES on ELT target $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$10, and achieving $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$11 would open direct acceleration mapping via globular clusters. Synergies proposed for the Milky Way include Gaia proper-motion data for perspective corrections, joint Bayesian analysis combining rotation curves, LOS accelerations, and stellar-kinematic data, and extension to other targets such as pulsars, eclipsing binaries, and gravitational-wave line-of-sight drift (Barbosa et al., 30 Jun 2025).

In gravitational-wave astronomy, projected precision improves strongly with low-frequency reach and signal duration. For BNS-like systems at $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$12, A+ forecasts give $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$13, Cosmic Explorer or Einstein Telescope reach $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$14, and DECIGO reaches $$\sigma_a \lesssim 0.6\,\mathrm{cm\,s^{-1}\,decade^{-1}}$$15. This suggests that routine LOSA detections are unlikely with current ground-based sensitivity but become plausible in third-generation and multiband observations, where the long inspiral can accumulate measurable Doppler dephasing (Vijaykumar et al., 2023).

Taken together, these developments establish line-of-sight acceleration as a unifying observable across Galactic dynamics, pulsar timing, and gravitational-wave astronomy. Its defining feature is operational rather than domain-specific: it is a direct measurement of acceleration along the observer’s line of sight, and therefore a local probe of the gravitational field or of environmental forcing. The major technical challenge is not conceptual identifiability but precision and waveform or calibration control. This suggests that the long-term significance of LOS acceleration will depend on whether large samples and physically complete forward models can outpace the systematics that presently dominate subleading Doppler effects.