Gravitational Acceleration Model

• Gravitational acceleration models are frameworks that map physical configurations to measurable acceleration or phase shifts, integrating methods from quantum interferometry to modified dynamics.
• They employ forward laws and inverse models—including likelihoods, neural networks, and Kalman filters—to deduce g from astronomical, metrological, and astrophysical data.
• These models drive advances in precision gravimetry, galactic rotation curve analysis, and cosmological acceleration while addressing regime limitations and compatibility with established tests.

Searching arXiv for recent and relevant papers on gravitational acceleration models. In current arXiv usage, a gravitational acceleration model is not a single canonical formalism but a family of constructions that predict, modify, or estimate gravitational acceleration from specified source distributions, spacetime backgrounds, or data streams. The literature includes relativistic phase models for optical and gravitational-wave signals, modified-gravity laws for galactic and cosmological dynamics, astrophysical acceleration models for spectroscopic anomalies, and inferential schemes that reconstruct local $g$ or $\nabla\Phi$ from stellar, pendular, atomic, or Kibble-balance measurements (Chen et al., 2018, Green et al., 2019, Kipper et al., 2018, Liu et al., 2024).

1. Scope and formal structure

Across the papers considered here, the common mathematical core is an explicit map from physical configuration to acceleration or to an acceleration-sensitive observable. In some cases the model is a forward law, such as a modified radial force or a phase shift. In others it is an inverse model, in which a likelihood, filter, or trained network returns an estimate of $g$ from measurements.

Use	Representative relation	Papers
Quantum-optical gravimetry	$\psi=\omega g(HL+H^2)/c^3$	(Chen et al., 2018)
Astrophysical line-shift modeling	$\Delta v_i=a_{i,\parallel}\Delta t$	(Chen et al., 6 Jul 2025)
Modified-gravity dynamics	$$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$; $g(r)=-GM/r^2-a_R+(\Lambda/3)r$	(Green et al., 2019, Grumiller, 2010)
Local field inference	$a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$	(Kipper et al., 2018)
Precision metrology	$g=4\pi^2L_{\rm eff}/T^2$; Kalman and transfer models for local $g$	(Prusty et al., 30 Mar 2025, Jiang et al., 2022, Liu et al., 2024)

A useful distinction is between state laws, which prescribe how gravity acts, and measurement laws, which prescribe how gravity enters an observable. The former include Newtonian extensions, MOND-like interpolations, Rindler terms, scalar-tensor-vector accelerations, and tensor-mode backreaction. The latter include interferometric phase accumulation, line-of-sight Doppler drift, orbital-arc likelihoods, and metrological transfer corrections.

2. Relativistic phase and waveform models

A particularly direct gravitational-acceleration model is the optical-fountain interferometer of Chen and Ralph, where only one arm of a Mach-Zehnder experiences a change in gravitational potential. Working on a Schwarzschild background, the model yields a proper-time difference

$\nabla\Phi$0

and therefore a gravitationally induced phase

$\nabla\Phi$1

This converts estimation of $\nabla\Phi$2 into interferometric phase metrology. In the small-phase Mach-Zehnder limit, the shot-noise-limited phase sensitivity is $\nabla\Phi$3, and the paper defines a corresponding standard quantum limit for $\nabla\Phi$4 (Chen et al., 2018).

The same work extends the passive Mach-Zehnder architecture in two directions. First, injecting a single-mode squeezed vacuum changes the lossless sensitivity by a factor $\nabla\Phi$5. Second, replacing the beam splitters by parametric amplifiers yields a two-mode SU(1,1) interferometer with altered loss sensitivity. Numerically, when detection loss is larger than $\nabla\Phi$6, the SU(1,1) interferometer outperforms the Mach-Zehnder with single-mode squeezing input; for currently available detector efficiency $\nabla\Phi$7 and low internal loss, the squeezed Mach-Zehnder remains the simplest, most sensitive choice. For $\nabla\Phi$8, $\nabla\Phi$9, $g$0--$g$1, and $g$2, the SQL gives $g$3 for $g$4 and $g$5 for $g$6. The same analysis states that reaching atom-fountain precision, $g$7, would require unrealistically large photon flux or arm lengths of many kilometres (Chen et al., 2018).

An analogous measurement-law perspective appears in accelerating compact binaries. Lazarow, Leslie, and Dai introduce a constant line-of-sight center-of-mass acceleration $g$8 into the TaylorF2 inspiral waveform through $g$9, obtaining a frequency-domain phase correction

$\psi=\omega g(HL+H^2)/c^3$0

The effect is strongest in the low-frequency part of the signal, so detectors with good low-frequency sensitivity are favored. A Fisher calculation gives, at signal-to-noise ratio $\psi=\omega g(HL+H^2)/c^3$1, detectability at Advanced LIGO A+ for $\psi=\omega g(HL+H^2)/c^3$2 in low-mass binaries, while Cosmic Explorer and Einstein Telescope can reach $\psi=\omega g(HL+H^2)/c^3$3--$\psi=\omega g(HL+H^2)/c^3$4. The linearized phase model remains valid up to $\psi=\omega g(HL+H^2)/c^3$5 for a light neutron-star binary with $\psi=\omega g(HL+H^2)/c^3$6 (Lazarow et al., 2024).

3. Modified-gravity laws and cosmological acceleration

Several papers formulate gravitational acceleration as an effective law containing additional fields or large-distance terms. In the geometric sigma model with four scalar fields, the nonrelativistic acceleration around a point source in de Sitter background takes the form

$\psi=\omega g(HL+H^2)/c^3$7

The four dominant contributions are identified as Newtonian, MOND-like, $\psi=\omega g(HL+H^2)/c^3$8CDM-like, and an exponentially growing repulsive term. The last becomes relevant only at scales beyond the observable Universe, while the time dependence is oscillatory and too slow to be directly observable today (Vasilic, 2018).

A different large-distance construction is the effective spherically symmetric model of Grumiller, for which

$\psi=\omega g(HL+H^2)/c^3$9

The novel term is the constant Rindler acceleration $\Delta v_i=a_{i,\parallel}\Delta t$0. The paper notes typical phenomenological values near $\Delta v_i=a_{i,\parallel}\Delta t$1 for galactic fits and the Pioneer anomaly, but also states that solar-system bounds require any universal $\Delta v_i=a_{i,\parallel}\Delta t$2 at $\Delta v_i=a_{i,\parallel}\Delta t$3, creating a scale-tension unless $\Delta v_i=a_{i,\parallel}\Delta t$4 depends mildly on system size (Grumiller, 2010).

In $\Delta v_i=a_{i,\parallel}\Delta t$5-gravity, the fundamental fields are the usual metric $\Delta v_i=a_{i,\parallel}\Delta t$6 and a second symmetric tensor $\Delta v_i=a_{i,\parallel}\Delta t$7. Massive particles do not follow a geodesic of $\Delta v_i=a_{i,\parallel}\Delta t$8 alone, while massless particles follow null geodesics of the effective metric $\Delta v_i=a_{i,\parallel}\Delta t$9. In the Newtonian limit,

$$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$0

Outside matter, the theory coincides exactly with General Relativity, and in cosmology it produces accelerated expansion without a cosmological constant. The paper quotes the bound $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$1 from gravitational red-shift tests (Alfaro, 2010).

A fourth route keeps General Relativity but attributes late-time acceleration to long-wavelength tensor modes in a closed Bianchi IX universe. In this model, the deviations from closed FRW are carried entirely by $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$2 gravitational-wave modes, and the second-order backreaction can make the average deceleration parameter negative. The paper reports that for growing-mode amplitudes of order unity, one obtains $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$3 without a cosmological constant. The same framework predicts a quadrupolar CMB modulation, low-$$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$4 alignment, and requires a closed cosmology with suppressed decaying modes, $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$5 compared with $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$6 (Schluessel, 2011).

4. Galactic and astrophysical phenomenology

Not all gravitational acceleration models are universal force laws; some are local astrophysical mechanisms. Chen et al. propose an oversimplified but potentially effective gravitational acceleration model for double-peaked narrow emission lines shifted in the same direction in kpc-scale dual-core mergers. The main galaxy hosts two narrow-line regions, one on the near side and one on the far side, while a companion without emission lines exerts an additional pull. Along the line joining the galaxies,

$$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$7

and the observed Doppler shifts are obtained from $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$8. In Monte Carlo experiments with $$g_{\rm MOG}(r)=g_N[1+\alpha-\alpha e^{-\mu r}(1+\mu r)]$$9, $g(r)=-GM/r^2-a_R+(\Lambda/3)r$0 runs satisfy both same-sign projected acceleration and $g(r)=-GM/r^2-a_R+(\Lambda/3)r$1, giving a probability of $g(r)=-GM/r^2-a_R+(\Lambda/3)r$2. The same paper identifies SDSS J001050.52-103246.6 as a plausible case, with $g(r)=-GM/r^2-a_R+(\Lambda/3)r$3 and $g(r)=-GM/r^2-a_R+(\Lambda/3)r$4, both blueshifted (Chen et al., 6 Jul 2025).

A laboratory-to-galaxy bridge appears in the differential MOND analysis of recent $g(r)=-GM/r^2-a_R+(\Lambda/3)r$5 experiments. The model assumes that MOND-type corrections depend only on the relative gravitational acceleration between a test mass and the array of source masses. For two bodies,

$g(r)=-GM/r^2-a_R+(\Lambda/3)r$6

Using the Quinn and Speake versus Schlamminger discrepancy, $g(r)=-GM/r^2-a_R+(\Lambda/3)r$7, at $g(r)=-GM/r^2-a_R+(\Lambda/3)r$8, the paper infers $g(r)=-GM/r^2-a_R+(\Lambda/3)r$9. It further states that the resulting curve closely traces galaxy rotation-curve data over $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$0 decades in $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$1 (Klein, 2015).

Trippe’s galactic-scale treatment adopts ad hoc massive gravitons and forbids graviton-graviton self-interaction. The resulting density profile is $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$2, the enclosed mass is

$a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$3

and the acceleration relation becomes

$a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$4

The model reproduces flat rotation curves, Tully-Fisher/Faber-Jackson, the mass discrepancy-acceleration relation, the surface brightness-acceleration relation, cluster kinematics, and Renzo’s rule without dark matter, but the paper explicitly describes it as a toy model rather than a consistent theory of gravitation. Its acceleration scale is quoted as $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$5, or $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$6 (Trippe, 2012).

Scalar-Tensor-Vector Gravity, or MOG, provides a covariant modified-acceleration law in the weak-field limit:

$a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$7

The parameters run with baryonic mass according to

$a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$8

with $a_j(\mathbf x)=a_{0,j}+b_j(x_j-x_{0,j})$9, $g=4\pi^2L_{\rm eff}/T^2$0, and $g=4\pi^2L_{\rm eff}/T^2$1. Using the SPARC sample and allowing galaxy parameters to vary within $g=4\pi^2L_{\rm eff}/T^2$2, the paper reports a dispersion of $g=4\pi^2L_{\rm eff}/T^2$3 of $g=4\pi^2L_{\rm eff}/T^2$4 dex and an empirical radial-acceleration scale $g=4\pi^2L_{\rm eff}/T^2$5 (Green et al., 2019).

5. Geodesy, planetary variation, and local field inference

In Galactic dynamics, Kipper, Tempel, and Tenjes model the local acceleration field directly rather than assuming a global potential. Within a small region $g=4\pi^2L_{\rm eff}/T^2$6, each Cartesian component is approximated by

$g=4\pi^2L_{\rm eff}/T^2$7

and each stellar orbit is integrated forward and backward through the box. The central assumption is that stars are phase mixed within the region, so the probability along each orbital arc is uniform in time. A kernel-smoothed phase-space density is then combined into a likelihood and optimized with MultiNest. On Gaia-Challenge mock data, the median recovery of the planar acceleration is $g=4\pi^2L_{\rm eff}/T^2$8--$g=4\pi^2L_{\rm eff}/T^2$9, the directional misalignment is $g$0 in the axisymmetric case and $g$1 in the barred case, and $g$2 stars per box are needed for $g$3 accuracy (Kipper et al., 2018).

A classical metrological branch models $g$4 through pendular dynamics. In the World Pendulum Alliance, the textbook relation

$g$5

is augmented by the finite-moment-of-inertia correction

$g$6

Using large $g$7 values, typically $g$8--$g$9, and multiple repetitions, the network reports measured values across fourteen stations that agree with the GRS 80 normal-gravity prediction to within $\nabla\Phi$00. The data also display the expected systematic depression at high-altitude Bogotá sites (Torres-Payoma et al., 2022).

A data-driven extension of the compound-pendulum experiment replaces direct analytical inversion by a neural predictor. The ANN uses three inputs—initial angle $\nabla\Phi$01, effective length $\nabla\Phi$02, and measured period $\nabla\Phi$03—with a $\nabla\Phi$04-$\nabla\Phi$05-$\nabla\Phi$06 architecture, $\nabla\Phi$07 hidden activation, linear output, min-max normalization, and Levenberg-Marquardt optimization. The best validation performance is reported at epoch $\nabla\Phi$08 with $\nabla\Phi$09. The traditional experiment gives $\nabla\Phi$10, while the ANN gives $\nabla\Phi$11 and a test-set mean absolute error of approximately $\nabla\Phi$12. The same paper notes overfitting beyond about $\nabla\Phi$13 hidden nodes (Prusty et al., 30 Mar 2025).

High-precision gravimetry uses explicit stochastic and transfer models. In a static atomic gravimeter, a two-state Kalman model removes short-term white phase noise and yields a residual Allan deviation of $\nabla\Phi$14 for more than $\nabla\Phi$15, corresponding to $\nabla\Phi$16 at a single-sample interval of $\nabla\Phi$17, even without seismometer correction (Jiang et al., 2022). In the Tsinghua tabletop Kibble balance, blind transfer of absolute gravity with relative gravimeters is handled by a sixth-order polynomial fit with optimal parameters $\nabla\Phi$18 and $\nabla\Phi$19, followed by horizontal and vertical gravity-gradient mapping, tide correction, and self-attraction modeling. The final determination at the mass position has an uncertainty of $\nabla\Phi$20 with $\nabla\Phi$21 (Liu et al., 2024).

Planetary applications treat gravitational acceleration as a latitude-dependent field on an oblate spheroid rather than a scalar constant. First-order Clairaut theory gives

$\nabla\Phi$22

while the third-order Cook expansion introduces $\nabla\Phi$23 and $\nabla\Phi$24 terms. Third-order corrections are negligible for Earth but significant for Jupiter, Saturn, and Jovian exoplanets. The associated asphericity-driven geostrophic wind is estimated to be $\nabla\Phi$25 for Earth and Jupiter, a few $\nabla\Phi$26 for Saturn, $\nabla\Phi$27 for a white-dwarf habitable-zone planet, and $\nabla\Phi$28 for WASP-19b (Haqq-Misra et al., 2016).

6. Regimes of validity, limitations, and recurrent controversies

The first recurrent issue is terminological. The phrase gravitational acceleration model spans incompatible objects: a quantum-optical phase transfer function, a modified dynamical law, a Monte Carlo mechanism for line shifts, and an estimator of local $\nabla\Phi$29. This suggests that the term is best treated as a category label rather than the name of a single theory.

The second issue is model status. Some proposals are explicitly limited in scope. The DPNEL merger model is described as oversimplified and neglects NLR-NLR gravity, gas or dynamical friction, radiation pressure, and changing NLR mass distribution; it is intended only for kpc-scale dual cores in early merger stages (Chen et al., 6 Jul 2025). Trippe’s galactic-scale scheme is likewise ad hoc and is explicitly not yet a consistent theory of gravitation (Trippe, 2012).

A third issue is regime dependence. The optical-fountain interferometer is based on present technology, but its own performance estimates state that atom-fountain precision would demand unrealistically large photon flux or many-kilometre arms; SU(1,1) superiority appears only above the quoted detection-loss threshold, whereas low internal loss and $\nabla\Phi$30 favor the squeezed Mach-Zehnder (Chen et al., 2018). The orbital-arc inference method requires local phase mixing and sufficiently populated regions, so its validity is tied to box size and stellar sampling density (Kipper et al., 2018). For oblate planets, first-order gravity is adequate when $\nabla\Phi$31 and $\nabla\Phi$32 are small, but third-order terms become necessary once either parameter exceeds roughly $\nabla\Phi$33 (Haqq-Misra et al., 2016). In accelerating-binary waveforms, the first-order treatment in $\nabla\Phi$34 breaks down above $\nabla\Phi$35 (Lazarow et al., 2024).

A fourth issue concerns compatibility with established tests. The large-distance Rindler model gains phenomenological leverage from the constant term $\nabla\Phi$36, but the same paper notes solar-system bounds much smaller than the values suggested by galactic fits (Grumiller, 2010). By contrast, $\nabla\Phi$37-gravity is constructed so that outside matter it coincides exactly with General Relativity, and its departures appear in the Newtonian rescaling and cosmological sector (Alfaro, 2010). The Bianchi IX tensor-backreaction model retains General Relativity but requires a closed cosmology and finely suppressed decaying modes to remain compatible with CMB isotropy (Schluessel, 2011). MOG avoids an arbitrary interpolation function but replaces it with mass-dependent running parameters $\nabla\Phi$38 and $\nabla\Phi$39 (Green et al., 2019).

Taken together, these works show that gravitational acceleration modeling is a cross-domain enterprise rather than a single program. In one branch, gravity is encoded as a measurable phase, timing, or frequency perturbation; in another, it is rewritten as an effective law with extra fields or scale-dependent corrections; in a third, it is inferred statistically from trajectories or metrological data. The technical unity lies not in shared ontology but in the requirement that acceleration be made operationally calculable from geometry, dynamics, or observation.