WorksPhysics and Astronomy
Direct inference of cosmological mass growth via timing of black hole - pulsar binaries
Abstract
Cosmologically coupled black holes (CCBHs) respond to the expansion of the Universe, independently of conventional astrophysical processes. Recent CCBH solutions in General Relativity demonstrate evolution of gravitating mass , where is the cosmological scale factor. This can be tested through the orbital period timing of black hole (BH) pulsar (PSR) binaries. In contrast to previous methods used to constrain , pulsar timing measurement baselines span decades instead of gigayears, allowing sources of bias to be well-controlled. As a test case, we analyze the globular cluster BH+PSR candidate J0514–4002E (spin , orbital period ) under the assumption that the companion is a CCBH. Taking the difference between the observed orbital and spin period derivatives removes all kinematic effects, leaving only the pulsar's intrinsic spin-down as unknown. Setting , its physical minimum, gives a robust upper bound at confidence. We estimate through the population of millisecond pulsars matched in spin and infer and , using and , respectively. Adopting typical baseline scaling of timing precision, the value , argued elsewhere to link energized vacuum BH interiors to accelerated cosmological expansion, can be excluded or detected for this companion at and confidence by 2031 and 2036, respectively.
Introduction
Binary pulsars continue to provide the most precise tests of relativistic orbital dynamics (e.g. J. H. Taylor et al. [67]; M. Kramer et al. [51]). Predictions for these astrophysical systems are typically determined in the point-mass limit of post-Newtonian (PN) frameworks, expansions of gravitational dynamics in dimensionless velocity squared () (e.g. T. Damour & J. H. Taylor [26]; C. M. Will [69]). Timing measurements for binary systems can be done generically within metric theories of gravity, allowing all manner of novel General Relativity (GR) predictions, and possible departures from GR (e.g. N. Yunes et al. [70]), to be interrogated.
Cosmological coupling (CC) is a recent prediction in GR that highly localized regions of relativistic stress can (not must) influence global cosmological dynamics at leading order (e.g. K. S. Croker & J. L. Weiner [19]; K. S. Croker et al. [20]). The effect becomes relevant given the existence of non-singular and horizon-free BH solutions in GR (e.g. I. Dymnikova [28]; P. O. Mazur & E. Mottola [56]), first posited by E. B. Gliner [41] to form as regions of energized vacuum during stellar gravitational collapse. If BHs are non-singular and cosmologically coupled, they could drive accelerated cosmic expansion by transmuting baryons at stellar collapse into energized vacuum (e.g. D. Farrah et al. [34]; K. S. Croker et al. [18]). S. P. Ahlen et al. [2] have shown that the resulting cosmology mimics the primary observables of CDM, yet halves the Hubble tension and recovers a positive summed neutrino mass consistent with terrestrial and Lyman- constraints (I. Esteban et al. [31]; C. Cain et al. [14]; W. Elbers [30]; Y. Kageura et al. [47]).
The reciprocal effect, whether and how cosmologically coupled stress responds locally to cosmological dynamics, is a more nuanced question. In contrast to linear theories like electrodynamics, the existence of a solution in GR that satisfies cosmological boundary conditions does not guarantee uniqueness of that solution (C. M. Bender & S. A. Orszag [7]). Recent progress has instead focused on large classes of solutions. V. Faraoni & M. Rinaldi [32, 33] have established via geometric methods that the radii of BH apparant horizons must evolve toward comovement within a Robertson-Walker (RW) cosmology. Because radius often relates to gravitating mass in compact object solutions, a natural local CC phenomenology is to adjust the gravitating mass
so that averages over a population of CCd systems reproduce large-scale CC phenomenology (K. S. Croker et al. [21]). The existence of solutions, like those of F. Kottler [50]; G. C. McVittie [58]; A. Einstein & E. G. Straus [29]; R. Gaur & M. Visser [39], does not exclude 1. M. Cadoni et al. [13, 10, 12, 9] have constructed local solutions for BHs with . Moreover, K. S. Croker et al. [22] have constructed a point-mass model in linearized GR, where the Hernandez-Misner-Sharp (HMS) mass evolves with any permitted. Because the HMS mass is geometrical, orbital dynamics would respond observably to this growth.
Observational searches for purely cosmological mass growth are complicated by the slowness of cosmological effects over most of the universe’s history. Existing approaches can be broadly divided into three categories: differential population studies, BH systems with age proxies, and aggregate population studies. Differential studies of the SMBH populations of quiescent elliptical galaxies by D. Farrah et al. [35, 36], which carefully account for measurement and selection bias, find consistency with passive cosmological growth at . Inspiring SMBHs source the low frequency portion of the stochastic gravitational wave background and the anomalously large signal amplitude reported by NANOGrav (G. Agazie et al. [1]) could be a signal of growth in this population (M. Lacy et al. [52]; M. Calzà et al. [15]).
For lower-mass BHs, the situation is less clear. Stellar mass BHs in environments where reliable ages can be estimated, as well as the LIGO merging population, can be used to constrain once a minimum BH mass is asserted. Adopting gives from all these populations (K. S. Croker et al. [21]; C. L. Rodriguez [63]; S. Ghodla et al. [40]; R. Andrae & K. El-Badry [5]; E. Mlinar & T. Zwitter [60]) except possibly low-mass X-ray binaries (S.-J. Gao & X.-D. Li [38]). L. Amendola et al. [4] have pointed out that a lower-bound BH mass from stellar-collapse is a feature of Kerr-like singular solutions, however, and need not hold
Corresponding author: Kevin S. Croker
[email protected] in general for non-singular BHs. Even so, would imply birth masses are possible (although not mandatory, e.g. J. C. A. Miller-Jones et al. [59]). Finally, P. Shah et al. [65] find no increased scatter in supernovae magnitudes at due to strong gravitational lensing. Such a signature is expected from a uniformly distributed cosmologically coupled BH population abundant enough to drive accelerated expansion, but a consistent confrontation with data where CCBH lensing strength decreases backward in time has yet to be performed. To begin to make sense of this observational landscape, we require a direct probe of the phenomenon.
In this paper, we utilize the exquisite timing precision of pulsar binaries to perform the first direct inference of with the BH+PSR candidate J0514–4002E reported by E. D. Barr et al. [6]. From J0514–4002E, we extract a robust observational upper-bound for . We then leverage the larger Galactic field PSR population to present the first direct inference of . Given the existing baseline data, we highlight that can be confirmed or refuted in J0514–4002E at confidence within the next four years. In a comprehensive Appendix § 7, we generalize well-known results on PN orbital mechanics to the case of a single time-varying mass with a preferred frame (the Cosmic Microwave Background, CMB) via Hamiltonian methods and the Poisson bracket. For dimensional clarity, we explicitly indicate light speed .
Methods
Using data from the Green Bank Telescope and the Karoo Array Telescope (MeerKAT), E. D. Barr et al. [6] report timing observations of J0514–4002E, a 5.6 ms PSR in an eccentric binary orbit within GC NGC 1851. The system is noteworthy because the pulsar’s unseen companion’s mass is 2.09 to 2.71 at 95% confidence, placing the companion in the mass gap between known neutron stars and BHs. In Table 1, we provide the relevant orbital properties of J0514–4002E as determined by E. D. Barr et al. [6] via a global post-Keplerian fit. In §B.4-§B.6 we show that CC does not measurably impact the periastron precession , Einstein delay , and orthometric Shapiro amplitude used to extract these mass measurements from J0514–4002E.
| Parameter | Value | Note |
| Total mass | ||
| Pulsar mass | ||
| units | ||
| units | ||
| 53731 MJD 59793 MJD |
Table 1. PSR J0514–4002E Binary Orbital and Pulsar Parameters by E. D. Barr et al. [6]. Reported uncertainties are 68% confidence. Dots denote precision beyond what is required for computation in this work.
P. C. C. Freire et al. [37] gives the relation between observed and intrinsic orbital period derivatives for pulsar binaries in globular clusters (GCs),
Here is the composite proper motion, is the distance to the pulsar binary center-of-momentum (the Shklovskii term, often approximated via the distance to the GC center), is the line-of-sight acceleration due to the immediate gravitational environment of the system, and is the acceleration difference between the Solar system and the target GC within the galaxy field, projected along the direction to the cluster. J0514–4002E shows no evidence of material in the binary environment, so there is no orbit evolution from accretion or mass-loss.
We assume the companion is a cosmologically coupled BH, with far field physics as modeled by K. S. Croker et al. [22]. In §B.3, we show that the only additional measurable GR contribution (by 6 orders of magnitude) to orbital period evolution in BH+PSR binaries due to CC is,
per orbit.2 Here we have introduced the mass ratio,
where and are the instantaneous pulsar and companion (CCBH) masses, respectively, and is defined in Eq. (27).
As shown in §B.7, orbital period change from CC complements that from radiative losses, so that intrinsic orbital period evolution becomes,
We assume that no additional radiative degrees of freedom are at play.
Standard analyses assume , and thus . Because Shklovskii and Galactic accelerations are typically small, Eqs. (1) and (4) imply that any observed is naturally interpreted as an estimate of . Because pulsar spin derivative and orbital period derivative are both offset by the same secular Doppler shift, the kinematic contributions to are identical to those in Eq. (1),
An orbital estimate of can then be used to compute the intrinsic pulsar spin derivative , if the Shklovskii and Galactic terms are known.
If the BH is cosmologically coupled with unknown , however, cannot be extracted from the orbital timing of the GC binary because it is degenerate with in Eqs. (1) and (4). For the same reason, inference of , and thus , cannot be performed from orbital timing alone. We can resolve this degeneracy by inverting the traditional procedure and using the spin period derivative as a proxy for kinematic effects. Formally, we subtract Eq. (5) from Eqs. (1) and (4) then rearrange to find,
This relation is entirely between binary observables and quantities intrinsic to the binary orbit and the bound objects. In principle then, if can be constrained through other data, the observed derivative in spin period can be used to estimate the intrinsic derivative in orbital period with exactly zero kinematic contamination.
Calibrator Pulsar Selection
To determine an empirical prior on , the quantity which enters Eq. (6), we begin with all millisecond pulsars ( ms) from the Australia Telescope National Facility (ATNF) pulsar catalog3. We select only Galactic field pulsars so that in Eq. (5). Field pulsars are identified using the ATNF GC associations and the catalog of known pulsars in GCs4 maintained by P. C. C. Freire, which is based on the McMaster catalog of Galactic globular clusters (W. E. Harris [44, 45]). For field pulsars within binary orbits, we select systems with compact companions that cannot contribute mass outflows. Thus, there is no Jeans-type accretion contaminant. We then require fractional uncertainty in proper motion , to enable accurate estimation of and Galactic acceleration within Eq. (5). For each object, the acceleration difference is computed using the P. J. McMillan [57] Galactic model and the Shklovskii acceleration is computed using the proper motions and distances5 tabulated in the catalog. Combined with the reported and , Eq. (5) is used to compute .
The resulting sample of pulsars has at 68% confidence and is shown in Figure 1, light-blue. We next construct the sample matched to J0514−4002E in median period ms, giving pulsars. This sample has at 68% confidence and is shown in Figure 1, medium-blue. This sample determines our empirical prior on .
Figure 1. Distributions of from millisecond pulsars, selected from the ATNF catalog to be free of local accelerations. (Light blue, dotted) Field millisecond pulsars with proper motion measured to fractionally better than 10% and, if in a binary, have compact companion. (Dark blue, solid) The subsample () with median period matched to J0514−4002E, ms. This sample is adopted as a prior for the intrinsic period derivative of J0514−4002E.
The intrinsic spin derivatives computed for all calibrator pulsars are determined independently of any orbit. Thus, they are not affected by the presence or absence of CC. We exclude millisecond PSRs within GCs from the calibrator sample because estimation of from this population requires orbital timing. The inferred from would be contaminated by if the companion were actually a CCBH with . Inclusion of contaminated systems would bias away from its true value toward zero. GC pulsars with optically or spectroscopically confirmed white dwarf companions avoid this ambiguity, but we defer their inclusion to future work.
Results
Any pulsar without evidence of ongoing mass accretion (active recycling) must physically satisfy . From Eqs. (6) and (38), any physical decreases the inferred , because both terms contribute to with the same sign. Setting and adopting Planck from M. Tristram et al. [68] then gives a robust upper bound6 on ,
We display this exclusion in Figure 2. We next compute with Eqs. (38) and (B64) in draws from the parameters in Table 1 regarded as Gaussian (we symmetrize to ), and from the median-matched sample in Figure 1 with replacements, to find

Figure 2. Constraints on stellar-mass BH from BH+Pulsar candidate J0514–4002E, assuming Planck ([68]). Orbital period derivative, and thus , is determined for this globular cluster pulsar via the intrinsic spin prior matched in median pulse period to 5.6 ms, displayed in Figure 1. Values for that classically imply causality violation (grey, hatched) are indicated. Robust exclusion at confidence (purple) is independent of intrinsic spin.
where we also include from the most recent measurement by H0DN Collaboration et al. [43]. Both values are consistent with existing constraints on . The Dominant Energy Condition (DEC) requires for stressed causal material. Eqs. (7) and (8) do not yet discriminate any physical value of from 0. Instead, they are a consistency test on the method: there is no a priori constraint on and the method could have recovered values entirely incompatible with the DEC.
Radio pulsar timing uncertainties scale reliably with baseline observation. In §C, we compute the additional baseline required for a desired confident detection of some at some desired significance. The additional baselines computed from Eq. (C77), assuming detector cadence about halfway between continual and endpoint, are given in Table 2.
| Target | Exclude / Detect | Baseline End (50%) | Baseline End (95%) |
| 1 | January 2035 | 2046–2051 | |
| 1 | 2041–2043 | 2058– | |
| 1 | |||
| 3 | May 2023 | October 2028 | |
| 3 | November 2026 | July 2031 | |
| 3 | June 2032 | August 2036 |
Table 2. Timing baseline endpoints by which J0514–4002E alone excludes or detects literature-predicted . Assuming improvement in orbital period derivative precision with a timing baseline anchored in 2005, we report endpoints for 50% and 95% probability of achieving the target confidence. Limitations based on our current cleaned and median-matched field millisecond pulsar calibrator for indicated with . Median endpoint reported if uncertainty spans months, 68% confidence interval reported if uncertainty spans years (see §C).
Baseline of E. D. Barr et al. [6] ends in August 2022
To our knowledge, the most recent (proprietary) timing data are from January 2026. Our forecast motivates increased cadence.
Discussion
Our prior on the intrinsic period derivative, shown in Figure 1, could be improved through Bayesian population inference to include selection effects. This becomes more important once exclusion or detection of becomes possible for J0514−4002E (c.f., Table 2), and the uncertainty in the parameters of J0514−4002E begin to shrink closer to that of our currently adopted prior’s width. In this study, we have taken the prior to be a discrete set of observed period derivatives, while, in general, one should use the observed values to characterize the underlying population distribution for the intrinsic period derivative (e.g. J. Alsing et al. [3]). For example, one could assume that the true prior is Gaussian in with some mean and variance, then infer the mean and variance from the observed values. Doing so would incorporate the uncertainties from each millisecond pulsar system, as well as selection effects. By this, we mean that some portions of our sample of field millisecond pulsars could be over-represented (e.g. D. R. Lorimer [55]; L. V. Drummond et al. [27]). Working with the inferred underlying distribution of pulsar intrinsic period derivatives will enable more robust inference of in future work.
The Square kilometer Array Observatory (SKAO) offers exciting prospects for pulsar detection and timing precision in Phase 1 of the SKAO using two telescopes with SKA-Low (50-350 MHz) and SKA-Mid (350 MHz-15.4 GHz) initially with the AA* array assembly (SKA-Low 307 stations 78,592 antennae from 2030, SKA-Mid 144 dishes from 2032) evolving to the full AA4 array assembly (SKA-Low 512 stations 131,072 antennae, SKA-Mid 197 dishes). In light of the MeerKAT experience, models need updating, but based on population studies and models in E. F. Keane et al. [48], the AA* array assembly is predicted to detect 820 millisecond pulsars (440 with SKA-Low, 380 SKA-Mid) within 5 degrees of the Galactic plane, rising to over 1000 predicted for AA4 array assembly. This population has the potential to significantly strengthen the calibrator sample.
In addition, approximately 110 (AA*) to 140 (AA4) binary neutron star systems are predicted. SKA1-Low detections will be mostly restricted to the local sky due to luminosity and interstellar medium effects, plus those more distant (out to 10-15 kpc bringing local group galaxies into the frame) at higher Galactic latitudes. Pulsars in the latter group will likely be young with high luminosity and correlate with supernova remnants. Since the SKA will be able to perform acceleration searches (up to and including 350 m s) on all of the incoming pulsar search data, detection of possible BH+PSR systems in the Galactic disk becomes possible where the BH mass is at least 15 M (B. W. Stappers et al. [66]).
In order to translate the above considerations into an estimate of how many BH+PSR binaries the SKA may observe, we need a model for the population of such objects in the Galactic field. One such model that focuses on the formation of BH+PSR systems from isolated massive binary stars uses the binary population evolution synthesis code COMPAS and predicts the presence of 50–2000 binaries. Using mock-surveys, such as those done by R. S. Booth et al. [8] for MeerKAT and assumptions about the survey parameters (in particular the area of the sky surveyed and the integration time), D. Chattopadhyay et al. [16] predict that “the SKA will observe 1–80 PSR+BHs, with 0–4 binaries containing millisecond pulsars.” The uncertainty comes from feasible pulsar evolution, massive binary evolution, and in re-scaling their model to the Milky Way population (e.g. assumptions about the number of stars in the Milky Way, the binary fraction, and initial mass function).
In summary, at coupling strength , BH+PSR binaries receive an additional contribution to the rate of orbital period change . Orbital period timing of BH+PSR binaries in globular clusters is contaminated by the cluster acceleration, but kinematic effects can be removed given knowledge of the PSR intrinsic spin down. Asserting zero spin-down for the GC BH+PSR candidate J0514−4002E, the smallest physical value, gives a robust upper-bound at 95% confidence. Estimating this PSR spin down from a calibrator “gold sample” of Galactic field PSRs matched in spin period gives and depending on adopted . A definitive measurement of would provide striking evidence of the “scale breaking” possible within non-linear theories like GR, contributing deep insight into the astrophysical nature of BHs. Our results show that for J0514−4002E will be excluded or detected in 95% of measurements at the confidence level by 2031 (2036 for ). Near-future experiments such as the Square Kilometer Array will further strengthen the calibrator sample and are expected to observe BH+PSR binaries with pulse periods of milliseconds.
Acknowledgments
KC thanks David C. and Marzia C. Schainker for their hospitality during the preparation of this work and K. Schmidt (ASU) for discussion. GBB acknowledges support from JWST-AR-04695 and JWST-AR-10799, which were provided through a grant from the STScI under NASA contract NAS5-03127. CC acknowledges support from the Kavli Institute for Cosmology at Cambridge and the Beus Center for Cosmic Foundations at Arizona State University. Anthropic’s Claude (Opus 5.5) was used to locate pertinent references, debate analysis approach, verify longhand calculations, verify catalog pulls, and review drafts of this manuscript. All manuscript text, derivations, analysis code, and figures were produced manually by the authors. RAW acknowledges support from NASA JWST Interdisciplinary Scientist grants NAG5-12460, NNX14AN10G and 80NSSC18K0200 from GSFC. JWR acknowledges funding from U.S. Department of Energy Grant No. DE-SC0016021. NF acknowledges funding from the U.S. Department of Energy Grant No. DE-SC0010008. L.T. acknowledges support from JSPS Kakenhi grants 24K22878, 26K17136, JST PRESTO JP-MJPR26287514, and Royal Society International Collaboration Award Round 2 (Japan) 252140. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant Nos. 2139433 and 2233001. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
CMB Dependent Quantities
The coordinates and motion of J0514–4002E were assumed to be that of NGC 1851. The position and radial velocity from W. E. Harris [44, 45] were used along with the distance and proper motion from M. Libralato et al. [54]. The velocity of NGC 1851 with respect to the solar system barycenter (SSB) is, as IRCS cartesian vector, .
To account for the motion of the SSB relative to the CMB, the CMB dipole moment from Planck Collaboration et al. [62] was used, which points towards the Galactic coordinates . As a cartesian ICRS vector this is
The velocity of NGC 1851 with respect to the CMB was then computed as
which becomes . Its magnitude is
and its line-of-sight component is
Theory
The solution of K. S. Croker et al. [22] (CC26) demonstrates cosmological HMS mass evolution is permissible within GR. The HMS mass impacts geodesic motion of test particles. In order that the Equivalence Principle remain satisfied, the inertial mass of a CCBH must also have temporal evolution matched to the HMS mass. For numerical bounds in this section, we adopt the largest theoretically motivated .
Let and be the masses of the pulsar and companion, with velocities and , respectively. We begin with the Lagrangian given in T. Damour & N. Deruelle [23], Eqns. 2.1a-c),
where is the magnitude of the relative separation vector and encodes the 1PN corrections to the Newtonian Kepler problem. CC26 describes a cosmologically coupled point mass at rest with respect to an isotropic CMB, receiving isotropic radial cosmological flux. Assuming that J0514–4002E is a binary PSR+BH system, we cosmologically couple Eq. (13) by introducing cosmological scale-factor dependence to the companion mass such that
where is the Hubble rate and is the constant cosmological coupling strength. The absence of explicit momentum transfer terms in Eq. (13) then implies that the harmonic coordinate system used to write Eq. (13) supplies the (non-orthonormal) coordinate basis for the that is at rest with respect to the CMB. Because CC26 establishes that , we continue to regard as constant.
Cosmological Coupling as Hamiltonian Perturbation
Our goal is to import as many T. Damour & N. Deruelle [23] results as possible. We will achieve this by encoding the impact of cosmological coupling, to the required accuracy, as a perturbation to the center-of-momentum (CM) frame 1PN Hamiltonian . We will then compute the altered temporal dynamics via Poisson brackets against this perturbed Hamiltonian, accounting for explicit time-dependence.
We begin with the characterization of orbital motion in terms of binary relative separation
relative velocity , and center of rest mass position (CRM, which is not the 1PN center of total energy position),
In these variables, but with time-dependent , the 1PN Lagrangian becomes
where contains the 1PN corrections. Note that is the binary’s motion relative to the CMB, because we have not changed frames, only switched to relative variables. The terms in square braces give a Poincaré invariant model if mass is fixed. The terms with are not invariant. Let dots denote contributions from . The conjugate momenta for Eq. (17) are,
with conserved. The Hamiltonian follows from Legendre transformation,
The portion of the Hamiltonian in square braces is identical in form to that obtained from the Poincaré invariant portion of Eq. (17) at fixed mass, after transforming into the binary center of momentum frame. In addition, the term containing produces plus negligible corrections. To the required accuracy then, the center of momentum binary problem as considered by T. Damour & N. Deruelle [23] can be cosmologically coupled by the addition of a perturbing potential,7
where is the binary center-of-momentum velocity relative to the CMB. The full 1PN Hamiltonian with variable-mass then has the same form as the Poincaré invariant 1PN Hamiltonian, plus a perturbation,
This allows us to use expressions for the 1PN conserved energy and angular momentum per unit reduced mass, and respectively, with the variable mass substituted. From this point forward, overset dots will denote derivatives with respect to binary center of momentum time (not CMB time).
Before we can make predictions for data, we need to characterize additional kinematic contributions to , compute the additional intrinsic orbital period change due to CC, and determine how other observables used to infer the binary’s masses, like the rate of periastron advance and Einstein delay , evolve when .
Kinematic acceleration
Because is not constant, but Eq. (19) is conserved in the center-of-momentum frame,
The term in square braces contributes negligible corrections and a term that orbit-averages to zero at the same precision. Thus, Eq. (1) acquires an additional secular term,
where is the projection of along the line-of-sight given by Eq. (12). The term contributes . Note that is the relative acceleration of the SSB and pulsar CM, due to their positions within the Galactic gravitational potential. It cannot absorb Eq. (24), which would be present even in the absence of the galaxy.
Orbital period change
To compute the contribution to from CC, we start from the 1PN generalization of Kepler’s 3rd law given by T. Damour & N. Deruelle [23],
and the 1PN conserved energy at fixed mass in the center of momentum frame,
As we proceed, it will become apparent that the corrections need not be explicitly expanded. To track the order of terms appearing in subsequent calculations, we introduce the usual small dimensionless parameter,
Taking a logarithmic derivative of Eq. (25) with respect to time and averaging over the 0PN orbit , following P. C. Peters [61] gives,
where we have discarded terms . Evaluation of the total mass (and reduced mass) is straightforward,
To get , we use the Poisson bracket (e.g. H. Goldstein [42]),
leveraging linearity of the bracket and that . Next express in canonical variables,
to enable the evaluation of Eq. (30),
Although is not constant, its derivative introduces , so its time-evolution introduces negligible terms.
In general, to extract the slow Hubble expansion (via time-dependent mass) from orbit-averages,
Taylor expand and mass terms like about ,
and use Friedmann’s equations to replace and note always,
expand similarly for , and drop . Although the required average is of ’s logarithmic derivative, because is constant along the 0PN orbit, it commutes through the average of . The term then becomes a total derivative that vanishes upon orbit-average. Substituting Eq. (29) gives,
The Virial Theorem applied to the 0PN orbit gives
Substituting Eqs. (29), (36) and (37) into Eq. (28) finally gives
This is the Keplerian result with adiabatic invariants found by K. S. Croker et al. [17], generalized to a single cosmologically coupled mass and with an explicit error bound. Using the projected orbit semi-major axis reported by E. D. Barr et al. [6] to establish a conservative bound, the relative 1PN correction for J0514–4002E is,
which is beyond the current timing precision of by . We note that the term is of this leading order contribution.
The effect is always present, regardless of the value of . Note that is possible because, in the absence of horizons, the TOV upper-limit mass for NS stability does not set a lower-bound stellar collapse BH mass (e.g. L. Amendola et al. [4], Fig. 14).
Periastron precession
Knowledge of the mass ratio is required for accurate inference of from measurements. Measurement of the total mass is determined from periastron precession . Canonical angular momentum is defined
The Poisson bracket gives,
and so the orbital plane receives a small torque. To verify that the masses reported by E. D. Barr et al. [6] are not biased by CC, we obtain the periastron precession via the Laplace-Runge-Lenz vector,
Evaluating the Poisson bracket and adding the temporal partial gives,
where the notation encodes the deviation from the 1PN fixed-mass behavior. Within the square braces, the first term is constant under 0PN orbit average. The second term is the total derivative and vanishes under orbit average. The third term is most easily averaged in index notation,
Write the integrand as,
The symmetric portion is again and vanishes from the average. The antisymmetric portion can be recognized as , and so
The required average in the fourth term of Eq. (43) is,
T. Damour & N. Deruelle [23] give the relative separation and Jacobian between binary time and eccentric anomaly ,
where we have set the temporal and spatial eccentricities of the 1PN problem equal to the Keplerian eccentricity along the 0PN orbit. The fourth term average becomes,
where we have leveraged periodicity of to label the starting angle as 0. Let point from attractive focus to periastron. Then is the orbital plane basis8 in which the components of are given by S. A. Klioner [49],
At 0PN, the basis remains fixed and performing the integration gives . Assembling all the pieces gives,
A component of along indicates orbital stretch, because . The magnitude of the term driving CC precession is the overlap of the other term with ,
Here is the relative velocity at apastron, the orbital minimum . Now, note that the time-derivative of Eq. (42) can be written as,
and we find
The value is roughly smaller than the errors reported by E. D. Barr et al. [6]. We conclude9 that the extracted total mass is robust.
Einstein Delay
The Einstein delay gives the magnitude of the oscillation in pulsar clock rate relative to the binary CM clock rate (T. Damour & N. Deruelle [24], Eq. (20)),
Because the masses enter asymmetrically, measurement of can be used to determine the component masses when combined with the precession of the periastron . Take the logarithmic derivative, substitute Eqs. (29) and (B29), and 0PN orbit-average,
To resolve the eccentricity time evolution, take a logarithmic derivative of 0PN , and again substitute Eqs. (29) and (B29),
where we have extracted from the average because it is constant at 0PN. Integration of Eq. (51) within Eq. (41) over eccentric anomaly with Eq. (49) gives,
Assembling Eqs. (57) to (59) and compressing via standard Kepler relations gives
This can be put into more transparent form via the semi-latus rectum and semi-major axis relations given by L. D. Landau & E. M. Lifshitz [53], Eqs. 15.4,15.6 and Kepler’s 3rd law,
Using Eq. (37) and T. Damour & N. Deruelle [23], Eq. 3.6a, , where
So the relative shift due to CC is manifestly for J0514–4002E. We conclude that the component masses extracted with are robust.
Orbital inclination and Shapiro delay
The orbital plane is torqued because Eq. (41) is non-zero in general. This will change the orbital inclination angle to the line of sight . E. D. Barr et al. [6] use the sine of to develop bounds on , and trigonometric functions of enter computation of the orthometric Shapiro delay parameter , so we bound its evolution as well. Recall that is orthogonal to the orbital plane, and that . Regard as constant to determine the leading order CC impact and find,
Averaging over the 0PN orbit, substituting Eq. (59), and expressing in terms of gives,
The accumulated angle over a 25 year baseline at is at most . The reported which, at , is already uncertain. We conclude that inferences with orbital inclination angle and Shapiro delay are unaffected.
Extension of CC to higher PN orders
While we have worked only to 1PN order explicitly in this appendix because 1PN effects are currently what is measured and constrained for J0514–4002E, the Poisson bracket formalism immediately generalizes to higher PN orders. The essential point is that the frame-dependence of CC is encoded via the Hamiltonian perturbation Eq. (21), whereas the impact on explicitly time-dependent companion mass enters only through a temporal partial at fixed phase-space variables.
Explicitly, let be some generic phase-space function with explicit time-dependence. Evolved along the Hamiltonian flow, it is always true that,
Suppose the Hamiltonian is given through 2PN, i.e. it includes the textbook 1PN, 1.5PN, and 2PN pieces. Then the conserved through 2PN continues to commute with and so,
The conserved through 2PN also continues to commute with and so,
provided that CC does not alter the spin of the coupled object. Regardless of the PN order being interrogated, the surviving Poisson bracket always contributes frame-dependence corrections bounded above by relative to the decoupled effect, as we have just seen explicitly at 1PN. Meanwhile, the temporal partial contributions always introduce fractional corrections from explicit temporal mass evolution at , regardless of PN order.
Independence from Radiative Losses
Gravitational radiation in GR results from the time-varying quadrupole moment and enters at 2.5PN. In particular, for the decoupled problem, with taken from G. Schäfer & P. Jaranowski [64],
gives the usual P. C. Peters [61] radiative decay at fixed mass when averaged over the 0PN orbit,
because at fixed phase-space in the fixed-mass problem. As we have shown in B.3, the CC orbital decay is nothing but the additional pieces to this bracket,
which happen to be non-vanishing at much lower order. Of course, within this bracket, at 2.5PN acquires fractional corrections through . But the relative pattern of suppression for CC impacts remains unchanged. This is consistent with order-of-magnitude expectations from existing treatments of time-varying masses in the CM frame by A. M. Holgado & P. M. Ricker [46], when the variation time-scale is extremely slow compared to the orbit period.
For computational convenience, Eq. (B61) can be expressed in terms of using Eq. (25),
where is the chirp mass.
Forecasting
Model any realization of measurement after a baseline via a normal distribution with variance that depends on ,
Then, the probability of measuring any after baseline is
Suppose that we want to claim an or better detection of . Then the distance from zero of the measured must be at least , i.e.
Perform the substitution to bring Eq. (C66) into standard form,
where is unity ( integrated over ) minus the cumulative distribution function for the standard gaussian PDF. The 2 comes from even symmetry of the integrand. Our goal becomes to invert this relation,
and solve for the required to achieve some given confidence on our measurement. The linear error propagation formula applied to Eq. (38) (at fixed for simplicity, because uncertainty in is subdominant)
By far, the leading contribution is from . Solving for in this regime and substituting Eq. (38) for gives,
We apply linear error propagation to Eq. (6) and resolve in terms of the dominant uncertainties in observables,
Substitution into Eq. (C72) gives,
where we have leveraged the exquisite timing precision to regard and as constants that commute freely with . The uncertainty in is modeled analogously to the T. Damour & J. H. Taylor [26] analysis of PSR B1913+16,
Here and are constants determined by detector performance and observation cadence. Let be the E. D. Barr et al. [6] observation baseline. Then, given , is determined by the uncertainty in listed in Table 1,
Equating Eqs. (C70) and (C72), substituting Eqs. (C75) and (C76) and solving for relative baseline increase gives,
Finally, we compute Eq. (C77) over our parameter draws, and distinct realizations of the field pulsar calibrator sample (draws with repeats) to get estimates of . Confidence intervals at 16% and 84% for the forecast dates in Table 2 typically range from a few months for to a few years for , with increasingly asymmetric spread upward for high-significant forecasts.
References
- [1]Agazie, G., Anumarlapudi, A., Archibald, A. M., et al. 2023, ApJL, 951, L8, doi: 10.3847/2041-8213/acdac6DOI
- [2]Ahlen, S. P., Aviles, A., Cartwright, B., et al. 2025, PRL, 135, 081003, doi: 10.1103/yb2k-kn7hDOI
- [3]Alsing, J., Silva, H. O., & Berti, E. 2018, MNRAS, 478, 1377, doi: 10.1093/mnras/sty1065DOI
- [4]Amendola, L., Rodrigues, D. C., Kumar, S., & Quartin, M. 2024, MNRAS, doi: 10.1093/mnras/stae143DOI
- [5]Andrae, R., & El-Badry, K. 2023, A&A, 673, L10, doi: 10.1051/0004-6361/202346350DOI
- [6]Barr, E. D., Dutta, A., Freire, P. C. C., et al. 2024, Science, 383, 275, doi: 10.1126/science.adg3005DOI
- [7]Bender, C. M., & Orszag, S. A. 1978, Advanced Mathematical Methods for Scientists and Engineers
- [8]Booth, R. S., de Blok, W. J. G., Jonas, J. L., & Fanaroff, B. 2009, https://arxiv.org/abs/0910.2935
- [9]Cadoni, M., de Lima, L., Pitzalis, M., Rodrigues, D. C., & Sanna, A. P. 2026, arXiv e-prints, arXiv:2601.03296, doi: 10.48550/arXiv.2601.03296DOI
- [10]Cadoni, M., Murgia, R., Pitzalis, M., & Sanna, A. P. 2024a, JCAP, 2024, 026, doi: 10.1088/1475-7516/2024/03/026DOI
- [12]Cadoni, M., Pitzalis, M., & Sanna, A. P. 2025, JCAP, 2025, 051, doi: 10.1088/1475-7516/2025/02/051DOI
- [13]Cadoni, M., Sanna, A. P., Pitzalis, M., et al. 2023, JCAP, 2023, 007, doi: 10.1088/1475-7516/2023/11/007DOI
- [14]Cain, C., Van Engelen, A., Croker, K. S., et al. 2025, ApJL, 987, L29, doi: 10.3847/2041-8213/ade141DOI
- [15]Calzà, M., Gianesello, F., Rinaldi, M., & Vagnozzi, S. 2024, Sci. Rep., 14, 31296, doi: 10.1038/s41598-024-82661-8DOI
- [16]Chattopadhyay, D., Stevenson, S., Hurley, J. R., Bailes, M., & Broekgaarden, F. 2021, Monthly Notices of the Royal Astronomical Society, 504, 3682, doi: 10.1093/mnras/stab3973
- [17]Croker, K. S., Nishimura, K. A., & Farrah, D. 2020, ApJ, 889, 115, doi: 10.3847/1538-4357/ab5affDOI
- [18]Croker, K. S., Tarlé, G., Ahlen, S. P., et al. 2024, JCAP, 2024, 094, doi: 10.1088/1475-7516/2024/10/094DOI
- [19]Croker, K. S., & Weiner, J. L. 2019, ApJ, 882, 19, doi: 10.3847/1538-4357/ab32daDOI
- [20]Croker, K. S., Weiner, J. L., & Farrah, D. 2022, PhRvD, 105, 084042, doi: 10.1103/PhysRevD.105.084042DOI
- [21]Croker, K. S., Zevin, M., Farrah, D., Nishimura, K. A., & Tarlé, G. 2021, ApJL, 921, L22, doi: 10.3847/2041-8213/ac2fadDOI
- [22]Croker, K. S., Cohon, J., Toth, V. T., et al. 2026, A Strong Cosmologically Coupled Mass Growth Solution in Linearized General Relativity, https://arxiv.org/abs/2609.38303
- [23]Damour, T., & Deruelle, N. 1985, Annales de L’Institut Henri Poincare Section (A) Physique Theorique, 43, 107
- [24]Damour, T., & Deruelle, N. 1986, Annales de L’Institut Henri Poincare Section (A) Physique Theorique, 44, 263
- [25]Damour, T., & Taylor, J. H. 1991, ApJ, 366, 501, doi: 10.1086/169585DOI
- [26]Damour, T., & Taylor, J. H. 1992, PhRvD, 45, 1840, doi: 10.1103/PhysRevD.45.1840DOI
- [27]Drummond, L. W., Chatzidnou, K., & Fonseca, E. 2026, https://arxiv.org/abs/2609.03157
- [28]Dymnikova, I. 2002, Classical and Quantum Gravity, 19, 725, doi: 10.1088/0264-9381/19/4/306
- [29]Einstein, A., & Straus, E. G. 1945, Reviews of Modern Physics, 17, 120, doi: 10.1103/RevModPhys.17.120
- [30]Elbers, W. 2025, arXiv e-prints, arXiv:2508.21069, doi: 10.48550/arXiv.2508.21069arxiv.org/abs/2508.21069
- [31]Esteban, I., Gonzalez-Garcia, M. C., Maltoni, M., et al. 2024, JHEP, 2024, 216, doi: 10.1007/JHEP12(2024)216DOI
- [32]Faraoni, V., & Rinaldi, M. 2024, PRD, 110, 063553, doi: 10.1103/PhysRevD.110.063553DOI
- [33]Faraoni, V., & Rinaldi, M. 2026, arXiv e-prints, arXiv:2608.21222, doi: 10.48550/arXiv.2608.21222arxiv.org/abs/2608.21222
- [34]Farrah, D., Croker, K. S., Zevin, M., et al. 2023a, ApJL, 944, L31, doi: 10.3847/2041-8213/acb704DOI
- [35]Farrah, D., Petty, S., Croker, K. S., et al. 2023b, ApJ, 943, 133, doi: 10.3847/1538-4357/acac2eDOI
- [36]Farrah, D., Engholm, A., Hatziminaoglou, E., et al. 2025, ApJ, 981, 71, doi: 10.3847/1538-4357/adb0c7DOI
- [37]Freire, P. C. C., Ridolfi, A., Kramer, M., et al. 2017, MNRAS, 471, 857, doi: 10.1093/mnras/stx1533DOI
- [38]Gao, S.-J., & Li, X.-D. 2023, ApJ, 956, 128, doi: 10.3847/1538-4357/ace890DOI
- [39]Gaur, R., & Visser, M. 2024, PRD, 110, 043529, doi: 10.1103/PhysRevD.110.043529DOI
- [40]Ghodla, S., Easther, R., Briël, M. M., & Eldridge, J. J. 2023, The Open Journal of Astrophysics, 6, 25, doi: 10.21105/astro.2306.08199
- [41]Gliner, E. B. 1966, Sov. J. Exp. Theor. Phys., 22, 378
- [42]Goldstein, H. 1950, Classical mechanics
- [43]HODN Collaboration, Casertano, S., Anand, G., et al. 2026, A&A, 708, A166, doi: 10.1051/0004-6361/202557993
- [44]Harris, W. E. 1996, AJ, 112, 1487, doi: 10.1086/118116DOI
- [45]Harris, W. E. 2010, arXiv e-prints, arXiv:1012.3224, doi: 10.48550/arXiv.1012.3224DOI
- [46]Holgado, A. M., & Ricker, P. M. 2019, ApJ, 882, 39, doi: 10.3847/1538-4357/ab3293DOI
- [47]Kageura, Y., Ouchi, M., Naokawa, F., et al. 2026a, ApJ, 1009, 42, doi: 10.3847/1538-4357/ae92faDOI
- [48]Keane, E. F., Graber, V., Levin, L., et al. 2025, The Open Journal of Astrophysics, 8, 54256, doi: 10.3233/OJC1.54256DOI
- [49]Klioner, S. A. 2016, arXiv e-prints, arXiv:1609.00915, doi: 10.48550/arXiv.1609.00915DOI
- [50]Kottler, F. 1918, Annalen der Physik, 361, 401, doi: 10.1002/andp.19183611402DOI
- [51]Kramer, M., Stairs, I. H., Manchester, R. N., et al. 2021, Physical Review X, 11, 041050, doi: 10.1103/PhysRevX.11.041050DOI
- [52]Lacy, M., Engholm, A., Farrah, D., & Ejercito, K. 2023, Accepted at ApJ, arXiv:2312.12344. https://arxiv.org/abs/2312.12344
- [53]Landau, L. D., & Lifshitz, E. M. 1969, Mechanics
- [54]Libralato, M., Bellini, A., Vesperini, E., et al. 2022, ApJ, 934, 150, doi: 10.3847/1538-4357/ac7727DOI
- [55]Lorimer, D. R. 2008, Living Reviews in Relativity, 11, 8, doi: 10.12942/lrr-2008-8
- [56]Mazur, P. O., & Mottola, E. 2015, Classical and Quantum Gravity, 32, 215024, doi: 10.1088/0264-9381/32/21/215024
- [57]McMillan, P. J. 2017, MNRAS, 465, 76, doi: 10.1093/mnras/stw2759DOI
- [58]McVittie, G. C. 1933, MNRAS, 93, 325, doi: 10.1093/mnras/93.325
- [59]Miller-Jones, J. C. A., Bahrmian, A., Orosz, J. A., et al. 2021, Science, 371, 1046, doi: 10.1126/science.abb3363
- [60]Mlinar, E., & Zwitter, T. 2024, MNRAS, doi: 10.1093/mnras/stae059DOI
- [61]Peters, P. C. 1964, Physical Review, 136, 1224, doi: 10.1103/PhysRev.136.B1224DOI
- [62]Planck Collaboration, Aghanim, N., Akrami, Y., et al. 2020, A&A, 641, A1, doi: 10.1051/0004-6361/201833880
- [63]Rodriguez, C. L. 2023, ApJL, 947, L12, doi: 10.3847/2041-8213/acc9b6DOI
- [64]Schäfer, G., & Jaranowski, P. 2024, Living Reviews in Relativity, 27, 2, doi: 10.1007/s41114-024-00048-7
- [65]Shah, P., Davis, T. M., Vincenzi, M., et al. 2025 MNRAS, 536, 946, doi: 10.1093/mnras/stae2614DOI
- [66]Stappers, B. W., Keane, E. F., Possenti, A., & Stairs, I. H. 2018, Philosophical Transactions of the Royal Society A, 376, Issue 2120, doi: 10.1098/rsta.2017.0293DOI
- [67]Taylor, J. H., Fowler, L. A., & McCulloch, P. M. 1979, Nature, 277, 437, doi: 10.1038/277437a0DOI
- [68]Tristram, M., Banday, A. J., Douspis, M., et al. 2024, A&A, 682, A37, doi: 10.1051/0004-6361/202348015DOI
- [69]Will, C. M. 2018, Theory and Experiment in Gravitational Physics, doi: 10.1017/9781316338612
- [70]Yunes, N., Siemens, X., & Yagi, K. 2025, Living Reviews in Relativity, 28, 3, doi: 10.1007/s41114-024-00054-9