WorksPhysics and Astronomy
Joint Bayesian Inference of Dense Matter and f(Q) Gravity with Nuclear and Neutron Star Constraints
Abstract
We investigate neutron star (NS) properties in symmetric teleparallel gravity while consistently accounting for uncertainties in the dense matter equation of state (EOS). We focus on the linear model and perform a joint Bayesian inference of the nuclear matter and modified gravity parameters using a Taylor expansion EOS around nuclear saturation density. The analysis incorporates complementary constraints from nuclear saturation properties, chiral effective field theory calculations of pure neutron matter, heavy-ion collision, and empirical nuclear data, the GW170817 tidal deformability posterior, and NICER mass–radius measurements. The resulting posterior distributions simultaneously constrain the dense matter and gravity sectors and quantify their correlations within a common framework. For a canonical NS, we obtain and , while the maximum mass is constrained to . The modified gravity parameters exhibit only weak direct correlations with most stellar observables, while a pronounced degeneracy remains within the gravity sector. Our results show that accounting for EOS uncertainties is essential for quantifying the interplay between dense matter physics and modified gravity and provide a systematic framework for confronting symmetric teleparallel gravity with nuclear and multimessenger NS observations.
Introduction
General relativity (GR) has passed every precision test from laboratory scales to astrophysical and cosmological observations [1]. Even so, open problems such as the late-time accelerated expansion of the Universe and the nature of dark energy [2, 3] have motivated a broad search for modified theories of gravity. The most familiar route keeps the curvature-based description and replaces the Ricci scalar in the Einstein-Hilbert action by a general function [4], at the cost of fourth-order field equations. Curvature, however, is not the only way to describe gravity. According to the geometrical trinity of gravity [5], GR can be rewritten equivalently in terms of torsion, as in the teleparallel equivalent of GR [6, 7], or in terms of non-metricity, as in the symmetric teleparallel equivalent of GR (STEGR) [8]. In STEGR, the connection is flat and torsion-free, and gravity is carried entirely by the non-metricity tensor . Although these three formulations give the same dynamics as GR, the equivalence is lost once they are extended to , or , so each extension can lead to physically distinct predictions [9]. Promoting the non-metricity scalar to an arbitrary function gives
gravity [8, 10]. Unlike gravity, gravity keeps the field equations for the metric second order. It has been studied widely in cosmology [11, 12] and increasingly for static, spherically symmetric and compact objects [13, 14, 15]. Neutron stars, with their strong gravitational fields and high central densities, provide a natural setting in which such modifications can be tested beyond cosmological scales.
The detection of gravitational waves (GWs) from the binary neutron star (BNS) merger GW170817 by the Advanced LIGO and Advanced Virgo detectors [16, 17], together with its electromagnetic counterpart [18], opened the era of multimessenger astronomy of neutron stars (NSs). During the inspiral, the tidal field of each star deforms its companion. The size of this deformation, quantified by the tidal deformability , depends on the internal structure of the star, and it leaves a measurable imprint on the GW phase. The signal therefore carries direct information about the NS equation of state (EOS) [19, 20, 21]. Complementary information comes from the Neutron Star Interior Composition Explorer (NICER). By modelling the X-ray pulses emitted from hot spots on the stellar surface, NICER has provided simultaneous mass and radius estimates for PSR J0030+0451 [22, 23], PSR J0740+6620 [24], PSR J0437−4715 [25] and PSR J0614−3329 [26]. Combining these multimessenger observations has led to increasingly tight constraints on NS properties and on dense matter [27]. However, almost all such analyses assume that GR is the correct theory of gravity. This assumption matters because the equilibrium structure of a NS is determined jointly by the EOS and by the theory of gravity. A change in one can be compensated by a change in the other, and the uncertainties from the two sectors currently overlap to a large extent [28]. The strong gravitational fields inside NSs therefore make it necessary to treat the uncertainties in the EOS and in the gravitational theory simultaneously rather than fixing either one.
The EOS of dense matter, i.e. the relation between pressure and energy density, remains the main source of uncertainty in modeling NSs [20]. Near the nuclear saturation density, , the EOS is reasonably well constrained. There, the energy per nucleon of asymmetric matter can be expanded in a Taylor series around . The expansion coefficients are the nuclear empirical parameters: the binding energy, incompressibility and skewness of symmetric matter, and the magnitude, slope and curvature of the symmetry energy [29]. The lower-order coefficients are fixed fairly well by finite nuclei, whereas the higher-order ones, which control the behaviour at larger densities, remain poorly known. Additional information comes from chiral effective field theory (EFT) calculations of pure neutron matter up to about – [30, 31], and from heavy-ion collisions and symmetry-energy measurements at supra-saturation densities [32, 33]. Beyond these densities the EOS can only be constrained indirectly through astrophysical observations. Bayesian frameworks that combine nuclear-physics inputs with NICER and GW data have therefore become the standard approach in GR [34, 35, 36]. In modified gravity, by contrast, NS studies usually adopt a single EOS or a small set of tabulated EOSs, and the gravity parameters are constrained conditional on that choice [37]. Two problems follow. First, the inferred gravity parameters absorb any mismatch between the chosen EOS and the true one. Second, many realistic EOSs have themselves been calibrated using astrophysical data interpreted within GR. A consistent test of gravity with NSs therefore requires an EOS description anchored in nuclear physics, whose parameters are inferred together with those of the gravity sector [28, 38].
In our previous work [37], we performed a Bayesian analysis of three models (linear, logarithmic, and exponential) using NICER mass–radius measurements, with the EOS fixed to DDME2. That analysis illustrated the sensitivity of the inferred gravity sector parameters to the assumed dense matter model. Since the EOS and the theory of gravity jointly determine the stellar structure, fixing the EOS can cause part of the model dependence to be absorbed into the inferred gravity parameters. In addition, many realistic EOSs have themselves been calibrated using astrophysical information interpreted within GR. A consistent test of gravity with NSs therefore requires an EOS description anchored in nuclear physics, whose parameters are inferred together with those of the gravity sector [28, 38]. More recently, GW170817 tidal information has also been employed to constrain NS configurations in modified gravity frameworks, including and nonminimally coupled models [39, 40]. These studies further illustrate the growing role of multimessenger observations in probing strong-field gravity with neutron stars. In this work, we perform a joint Bayesian inference of the dense matter EOS and the gravity sector within symmetric teleparallel gravity. For the matter sector, we adopt a Taylor expansion EOS around nuclear saturation density and treat the nuclear matter parameters as free quantities. For the gravity sector, we consider the linear model, which provides a consistent framework within the static, spherically symmetric coincident-gauge formulation adopted here. The analysis combines constraints from nuclear saturation properties, EFT calculations of pure neutron matter, heavy-ion collision and empirical nuclear data, the GW170817 tidal deformability posterior, and NICER mass–radius measurements. By sampling the dense matter and gravity parameters simultaneously, we quantify the correlations and degeneracies between the two sectors and determine their joint impact on neutron star properties.
The paper is organized as follows: Sec. II introduces the Taylor-expanded dense-matter EOS around nuclear saturation density. Sec. III describes the modified Tolman–Oppenheimer–Volkoff (TOV) equations that govern neutron star structure in gravity. The Bayesian framework used for the joint inference of the nuclear and gravity parameters is presented in Sec. IV. In Sec. V, we introduce the multiphysics data sets entering the likelihood, namely the EFT calculations of pure neutron matter, heavy-ion collision and empirical nuclear data, the GW170817 tidal-deformability posterior, and the NICER mass–radius measurements. Our results, including the posterior distributions of the nuclear and gravity parameters, their correlations, and the inferred neutron star properties, are discussed in Sec. VI. Finally, we summarize our findings and present concluding remarks in Sec. VII.
Equation of State of Dense Matter
We construct the dense matter equation of state using a Taylor expansion around the nuclear saturation density () [41, 42, 43, 44, 45, 46, 47]. The relevant nuclear matter parameters are treated as free parameters within the Bayesian analysis, allowing us to explore the density dependence of symmetric nuclear matter and the symmetry energy. Within the parabolic approximation, the energy per nucleon , at baryon density and isospin asymmetry , is expressed as
where is calculated using the -equilibrium and the charge neutrality conditions. The energy per particle for SNM and symmetry energy can be evaluated using nuclear matter parameters, and , which are defined as
where, (binding energy), , (incompressibility coefficient), (skewness), (kurtosis), and
where (symmetry energy coefficient), (slope parameter), (curvature parameter), (skewness), (kurtosis). At low densities, the core EOS is smoothly connected to the crust EOS. We adopt the Baym–Pethick–Sutherland (BPS) EOS [48] for the outer crust up to , while the core EOS is used above . In the intermediate density region, , the inner crust is described by a polytropic interpolation,
where the coefficients and are determined by requiring pressure continuity at the outer-crust–inner-crust and inner-crust–core boundaries. The resulting crust-core EOS should satisfy three essential conditions: (i) thermodynamic stability, (ii) causality (), and (iii) positive symmetry energy. This complete EOS provides the matter input for the stellar-structure equations in gravity, which are introduced in the following section.
Modified TOV Equations in Gravity
Throughout this work, we adopt the symmetric teleparallel formulation of gravity, in which the affine connection is flat and torsion-free and the gravitational dynamics are encoded in the nonmetricity scalar . Since the general formalism is well established, we do not reproduce it here and refer the reader to Refs. [8, 10] for its formulation and to Refs. [14, 37] for the conventions adopted in the present work. In particular, we work in the coincident gauge, , for which the nonmetricity scalar is determined entirely by the metric.
Field equations for a static, spherically symmetric star
The interior of a static, spherically symmetric star is described by the line element
with . The two metric potentials and carry the and components of the geometry, respectively. The stellar matter is treated as a perfect isotropic fluid [49, 50],
where and are the energy density and pressure, and the four-velocity satisfies .
For the ansatz (5), the non-metricity scalar depends only on the radial coordinate,
where a prime denotes . Substituting Eqs. (5)–(7) into the field equations yields the independent components
where , and . Equation (11) is off-diagonal and, on its own, it has no counterpart in GR. Its factor cannot be cancelled by anything else in a spherically symmetric configuration, so the equation must be satisfied by the remaining factors. For a non-linear model, , and the condition can only be met if , that is, if the non-metricity scalar is constant throughout the star. This is a direct consequence of the coincident gauge: fixing the connection to zero exhausts the gauge freedom, and the residual field equation then constrains the allowed configurations rather than being satisfied identically. The same point has been raised recently in analyses that keep the connection dynamical [51, 52].
Stellar structure equations
The conservation of the energy-momentum tensor gives the usual hydrostatic equilibrium condition,
Together with Eqs. (8)–(10), this closes the system. Eliminating and in favour of the metric functions, the structure of the star is governed by
Equations (12)–(14) are the modified TOV equations of gravity. Once a functional form of and an EOS relating to are specified, they are integrated outward from the centre subject to the regularity conditions
until the pressure vanishes, , which defines the stellar surface at radius . The exterior is vacuum, and for Eqs. (8) and (9) give
so the interior solution is matched to the Schwarzschild geometry by requiring at the surface.
In the limit one has and , so Eq. (11) is satisfied identically and Eqs. (8)–(10) reduce to the standard Einstein field equations. Departures from GR are therefore controlled entirely by the chosen functional form of . In this work we adopt the linear model,
The linear model is the only case in which the coincident gauge adopted here is internally consistent with : the off-diagonal equation (11) is satisfied identically, so the field equations impose no further restriction on and the spherically symmetric ansatz (5) can be used without over-constraining the solutions. This consistency comes at a definite price, which is worth making explicit. Substituting and into the covariant field equations, the derivative term proportional to vanishes and the remaining bracket reduces to a constant, , leaving
It is important to note that, in the linear model, and enter the field equations only through the product . Consequently, NS observables cannot independently determine these two parameters within the present formulation, leading to an intrinsic degeneracy between them. The individual marginalized posteriors of and should therefore be interpreted with this parameter degeneracy in mind, while the combination is the quantity directly relevant to the stellar-structure equations. The non-metricity sector therefore enters only through two constants: rescales the gravitational coupling, , while acts as an effective cosmological term in the stellar interior. General relativity is recovered for and . The connection does not appear in Eq. (18) at all, which is precisely why the gauge choice is harmless here, and it also means that the model does not probe non-metricity as a dynamical degree of freedom.
What the present analysis constrains is consequently a well-posed but specific question. Because is a single constant, common to the weak-field and strong-field regimes, variations in can be partially degenerate with changes in the EOS in stellar-structure observables. In particular, a smaller corresponds to a larger and therefore to a stronger effective gravitational coupling, which can be compensated by a stiffer EOS providing greater pressure support. The two effects can thus produce similar changes in the mass–radius relation, giving rise to an EOS–gravity degeneracy. Sampling jointly with the nuclear matter parameters measures directly how much of the inferred stellar structure can be reassigned from the matter sector to the gravitational sector, and how strongly the multiphysics constraints of § 5 break that reassignment.
Bayesian Analysis
We employ Bayesian inference to constrain simultaneously the parameters associated with the dense-matter EOS and the linear gravity model. We denote the complete set of model parameters by , which includes the nuclear matter parameters entering the Taylor-expanded EOS together with the parameters characterizing the model. For a given dataset , Bayes’ theorem gives the posterior distribution as
where denotes the likelihood of the data for a parameter set , while represents the corresponding prior distribution. The quantity is the Bayesian evidence and provides the normalization of the posterior. The resulting posterior distributions are used to determine the allowed parameter ranges and correlations between the dense-matter and modified-gravity sectors.
Likelihood
For a given parameter set , the likelihood measures the consistency of the model predictions with the adopted experimental and astrophysical constraints. Under the assumption that the datasets are independent, the joint likelihood is obtained by multiplying the individual contributions.
For an experimental quantity reported with a Gaussian uncertainty , we adopt
Here, represents the theoretical value corresponding to a given parameter set. We use this Gaussian form for the nuclear and terrestrial observables characterized by measured central values and uncertainties, including the saturation properties, pure neutron matter, and heavy-ion collision constraints.
For gravitational-wave observations, the likelihood depends on the component masses and their corresponding tidal deformabilities. For a given , the EOS and the gravity parameters uniquely determine the relation . The GW likelihood can therefore be written as
The mass distribution is taken to be uniform within the adopted mass interval,
Here and denote the lower and upper limits of the mass range considered for the GW event.
The NICER measurements provide joint posterior information on the neutron-star mass and radius. For each sampled parameter set, the corresponding mass–radius relation is obtained by solving the modified stellar-structure equations. The NICER likelihood is then evaluated as
For the X-ray likelihood, we take , while is determined by the maximum stable neutron-star mass predicted for the corresponding parameter set. The likelihoods from the different NICER pulsars are multiplied assuming that the measurements are statistically independent.
The complete likelihood employed in the present analysis is therefore written schematically as
where accounts for the microscopic pure-neutron-matter information, and incorporates the constraints obtained from heavy-ion collision experiments. The last two terms contain the gravitational-wave and NICER information, respectively. In this way, terrestrial nuclear information and astrophysical observations are included within a common Bayesian framework to constrain both the EOS and the parameters of linear gravity.
Multiphysics Data
In the present analysis, all nuclear, terrestrial, and astrophysical constraints are incorporated simultaneously within a common Bayesian framework. The adopted inputs include microscopic pure-neutron-matter calculations, empirical and heavy-ion collision constraints on nuclear matter, the GW170817 tidal-deformability posterior, and the available NICER mass–radius measurements. These complementary datasets probe different density regimes and therefore provide joint constraints on both the dense-matter EOS and the parameters of the linear model.
EFT
At sub-saturation and moderately suprasaturation densities, chiral effective field theory provides a controlled description of neutron-rich matter. In particular, LO calculations based on interactions constrained by nucleon–nucleon scattering, few-body observables, and nuclear saturation properties provide reliable predictions for pure neutron matter [53, 54, 55]. We use the pure-neutron-matter pressure, , obtained from the LO calculations up to a density of [54, 55].
Heavy-Ion Collision
We include empirical constraints on the symmetry energy , symmetry-energy pressure , and symmetric nuclear matter pressure . These constraints are inferred from nuclear masses, the neutron-skin thickness of , dipole polarizability, isobaric analog states, and heavy-ion collision measurements over the density interval – following the compilation given in Table I of Refs. [56, 57, 58].
GW170817
The tidal response of neutron stars in binary mergers provides an important probe of the high-density EOS. For GW170817, whose component masses lie approximately in the range 1.17–1.6 , gravitational-wave observations provide posterior information on the component tidal deformabilities [59]. We use the publicly available GW170817 posterior samples released by the LIGO–Virgo Collaboration5 as an astrophysical constraint in the Bayesian analysis. We note that these posterior samples were inferred using waveform models constructed within GR. In the present work, they are therefore employed as phenomenological constraints on the tidal deformabilities predicted within the adopted framework. A fully self-consistent treatment would require the binary dynamics and waveform model to be formulated directly in the corresponding modified-gravity theory.
NICER
Mass–radius measurements from NICER provide direct information on neutron-star structure through X-ray pulse-profile modeling. We include the publicly available posterior distributions for PSR J0030+0451 obtained independently by Riley et al. [60]6 and Miller et al. [61]7, together with the corresponding measurements of PSR J0740+6620 from Riley et al. [62]8 and Miller et al. [63]9. We further include the more recent NICER posteriors for PSR J0437-4715 from Choudhury et al. [64]10 and PSR J0614-3329 from Mauviard et al. [65]11. Together, these measurements provide coverage over a broad region of the neutron-star mass–radius plane and are incorporated jointly with the nuclear and gravitational-wave constraints.
Results and Discussion
In this section, we present the results of the joint Bayesian inference of the dense-matter EOS and the linear gravity model. The Taylor-expansion EOS is combined with the modified stellar-structure equations, and the nuclear matter and parameters are constrained simultaneously using the complete set of nuclear, heavy-ion, gravitational wave, and NICER data introduced in § 5. We first discuss the prior, posterior distributions and parameter correlations, and then examine the resulting constraints on the nuclear matter properties, the -equilibrated EOS, and NS properties, including the mass–radius relation, tidal deformability, and maximum mass.
Priors and Posterior distributions of parameters
The posterior distributions of the nuclear matter parameters (NMPs) and the linear model parameters are obtained by simultaneously imposing the nuclear, theoretical, and astrophysical constraints described in § 5. These include the basic properties of nuclear matter around saturation density, the pure-neutron-matter pressure predicted by LO chiral effective field theory, empirical information from heavy-ion collision experiments, and the GW170817 and NICER observations. The adopted prior distributions for the NMPs and the parameters are summarized in Table I. The saturation density and binding energy per nucleon are assigned Gaussian priors, while uniform priors are adopted for the remaining NMPs and the modified-gravity parameters , , and .
| min./ | 0.160 | -16.0 | 150 | -1500 | -5000 | 25 | 20 | -400 | -1500 | -5000 |
| max./ | 0.005 | 0.30 | 330 | 1500 | 5000 | 40 | 120 | 400 | 1500 | 5000 |
| min. | -15.0 | -5.0 | -1.0 | |||||||
| max. | 15.0 | 5.0 | 1.0 |
Table 1. Prior distributions of the NMPs. and are assumed as Gaussian distributions with a mean () and standard deviation (). All other NMPs are assumed to be uniformly distributed between a minimum (min.) and a maximum (max.) values as listed in the units of MeV. In below cells, we have listed the prior distributions of model parameters.
The joint posterior distribution of the nuclear matter parameters (NMPs) and the -gravity parameters is obtained from the product of the likelihood and prior distributions, as defined in Eq. (19). The marginalized posterior distribution of each parameter is obtained by integrating the joint posterior over all remaining model parameters. A posterior distribution that is appreciably narrower than its corresponding prior indicates that the parameter is constrained by the nuclear-theory and astrophysical data entering the likelihood. The one- and two-dimensional marginalized posterior distributions obtained for the Taylor-expanded EOS combined with the model are shown in Figure 1. The median values and the corresponding 68% credible intervals are reported above the diagonal panels, while the off-diagonal panels display the joint 68% and 95% credible regions.

Figure 1. The marginalized posterior distributions of the NMPs and model parameters, obtained through Bayesian inference.
The posterior distributions of several nuclear matter parameters (NMPs) are significantly narrower than their adopted priors, indicating that the combined nuclear and astrophysical constraints provide substantial information on the density dependence of the EOS. At the 68% credible level, we obtained , , , . The posteriors of , , , and are also appreciably constrained relative to their broad uniform priors, whereas remains largely prior dominated. For the linear model, the modified-gravity parameters are inferred as , , , with the posterior favoring , while the 68% credible intervals of both and include zero. The marginalized posteriors of and exhibit pronounced bimodal features, reflecting the intrinsic degeneracy associated with the fact that these parameters enter the stellar-structure equations only through the combination .
Several pronounced correlations are visible in the joint posterior distributions. The strongest anticorrelations occur between and , and between and , with , while and exhibit a strong positive correlation, . Other notable correlations include – (), – (), and – (). Moderate correlations, with –, are also found for –, –, –, –, –, and –. The strong correlation between and reflects the degeneracy associated with the constant contribution in the linear model. In contrast, the modified-gravity parameters are only weakly correlated with most NMPs, apart from moderate correlations of with () and ().
Posterior distributions of the EOS and NS properties
Using the posterior distributions of the model parameters shown in Figure 1, we reconstruct the density dependence of the nuclear and -equilibrated matter properties. Figure 2 shows the 95% and 68% credible intervals for (a) the symmetry-energy pressure, , (b) the pressure of symmetric nuclear matter, , (c) the symmetry energy, , and (d) the pressure of pure neutron matter, , as functions of the baryon density . The posterior bands are consistent with the experimental, empirical, and EFT constraints included in the analysis over the density region in which these data are available. The uncertainties remain relatively narrow around and moderately above the saturation density, while they gradually broaden toward higher densities, reflecting the weaker terrestrial constraints in this regime.

Figure 2. The 95% and 68 % credible interval plots for (a) the pressure of the symmetry energy, , (b) the pressure of SNM, , (c) the symmetry energy, , and (d) the pressure of PNM, , as a function of the baryon density , evaluated using the posterior distributions as listed in Figure 1. The corresponding experimental data incorporated into the Bayesian framework are also shown.

Figure 3. The 95% and 68% credible interval posterior distributions of the pressure of -equilibrated matter, (left), the proton fraction, (middle), and the squared sound speed, (right), as functions of the baryon density . The solid horizontal line indicates the nominal direct-Urca threshold .
Figure 3 presents the corresponding posterior distributions of the pressure of -equilibrated matter, proton fraction , and squared speed of sound . The pressure increases smoothly with density, as required for a thermodynamically stable EOS, while its uncertainty becomes appreciably larger at supranuclear densities. The proton fraction increases monotonically and crosses the nominal direct-Urca threshold, , at approximately for the median EOS. This suggests that sufficiently massive NSs may allow the direct Urca process, subject to their central densities and detailed particle composition. The squared sound speed increases with density, exhibits a relatively mild variation at intermediate densities, and rises again in the high-density regime. The median prediction remains well below the causal limit, , throughout the density range considered, while the credible intervals broaden substantially toward higher densities. Overall, these results show that the low- and intermediate-density behavior of the EOS is comparatively well constrained, whereas appreciable uncertainty remains in the supranuclear-density regime relevant to the cores of massive NSs.
The corresponding constraints on the macroscopic NS properties are shown in Figure 4. The left panel displays the 68% and 95% credible regions of the mass–radius relation together with the mass–radius constraints from the NICER observations and GW170817. The posterior favors a relatively narrow range of radii over a broad interval of stellar masses and remains compatible with the observational regions included in the Bayesian analysis. For a canonical NS, we obtained . The middle panel shows the tidal deformability as a function of stellar mass. As expected, decreases rapidly with increasing mass as the star becomes progressively more compact. The inferred value at is , consistent with the tidal deformability constraint from GW170817. The right panel presents the mass–compactness relation, where increases systematically with stellar mass. Overall, the posterior bands show that the combined nuclear and astrophysical constraints restrict the stellar radius, tidal response, and compactness within the linear framework.

Figure 4. The 95% and 68 % credible interval distributions for the radius, R(km) (left panel) and tidal deformability, (middle panel), and compactness (right panel) as a function of neutron star mass, . The astrophysical observations incorporated in the Bayesian framework are also shown.
To further quantify these results, Figure 5 shows the marginalized posterior distributions and correlations among selected NS properties, namely , , , , and . In addition to the constraints on and quoted above, we obtain and . The maximum mass is constrained to , indicating that the posterior EOS are compatible with the existence of massive neutron stars above the scale.

Figure 5. The marginalized posterior distributions of the NS properties, obtained through Bayesian inference.
Several correlations among these quantities are also evident from Figure 5. A strong positive correlation is found between and , reflecting the strong sensitivity of the tidal deformability to the stellar radius. In contrast, the central pressure is strongly anticorrelated with both and , with smaller radii and tidal deformabilities being associated with larger central pressures. The central squared sound speed shows broader anticorrelations with and and a positive association with . These correlations illustrate how the properties of dense matter are reflected in the macroscopic observables of neutron stars. By comparison, the correlations involving are considerably broader, consistent with its greater sensitivity to the high-density behavior of the EOS beyond the densities predominantly probed by a star.
Finally, to examine the interplay between the dense matter sector, the modified gravity parameters, and the resulting NS properties, we evaluate the Pearson correlation coefficients among all sampled parameters and selected NS observables. The corresponding correlation matrix is shown in Figure 6. Several characteristic correlation patterns are evident among the NMPs. The strongest anticorrelations are found between and and between and , both with . A strong anticorrelation is also observed between and (), while and

Figure 6. Pearson’s correlation matrix plots between the model parameters and selected NS properties.
are anticorrelated with . The saturation density shows strong positive correlations with () and (). These correlations indicate that the combined constraints restrict specific combinations of NMPs rather than acting independently on each parameter, reflecting their coupled role in determining the density dependence of the EOS.
The modified gravity sector exhibits a different correlation pattern. The strongest correlation is between and , with . This correlation reflects the intrinsic parameter degeneracy of the linear model, since and enter the stellar-structure equations only through the combination . Consequently, their correlated joint posterior and the bimodal features of the corresponding one-dimensional marginalized posteriors should be interpreted as manifestations of this non-identifiability rather than as evidence for distinct physical branches. In this sense, the combination is the quantity directly constrained by the NS properties, whereas and individually remain only weakly identifiable.
The parameter shows moderate correlations with () and (), while its correlations with the remaining NMPs are generally weak. Direct correlations between the modified gravity parameters and the NS properties are also small. In particular, and show negligible correlations with , , , , and , whereas exhibits only a weak positive correlation with () and weak correlations with the other stellar quantities. This behavior suggests that, within the linear framework, the stellar properties are predominantly shaped by the dense matter sector, with only a limited direct sensitivity to the modified gravity parameters over the posterior region explored.
Among the NS properties, the strongest correlation is found between and , with . The central pressure is strongly anticorrelated with both () and (). The central squared sound speed is moderately anticorrelated with () and (), while it is positively correlated with (). In contrast, shows only weak correlations with the canonical-star observables, with for , for , for , and for . The comparatively weak correlations involving are consistent with its greater sensitivity to the high-density EOS, which is less tightly constrained than the density region predominantly probed by a 1.4 star.
It is also instructive to compare the present results with our previous fixed-DDME2 analysis [37], in which the linear, logarithmic, and exponential models were investigated separately. For the linear model, the earlier analysis showed a strong correlation between and , while remained only weakly correlated with the other stellar observables. The gravity-sector parameters also exhibited a moderate correlation between and the characteristic nonmetricity scale, whereas was comparatively well constrained around the GR-like regime. These qualitative features persist in the present analysis when the EOS parameters are allowed to vary simultaneously with the parameters. In particular, and remain very strongly correlated (), while continues to show only weak correlations with the canonical-star observables. The correlation between and becomes more pronounced (), reflecting the degeneracy associated with their combined contribution to the linear model. At the same time, the direct correlations of the modified gravity parameters with the NS properties remain generally weak.
To further examine the impact of EOS uncertainties on the gravity sector, Figure 7 compares the one-dimensional marginalized posterior distributions of the linear parameters with those obtained in our previous fixed-DDME2 analysis [37]. The posteriors of and remain broad in both analyses, while the present calculation exhibits a more pronounced bimodal structure, consistent with the strong – correlation found in Figure 6. This behavior is a direct consequence of the intrinsic degeneracy between these parameters. As follows from Eq. (17), the linear model can be written as , so that and enter the stellar-structure equations only through their product .

Figure 7. Comparison of the one-dimensional marginalized posterior distributions of the linear parameters , , and , together with the derived combination , obtained in the present Taylor-expansion EOS analysis and in the previous fixed-DDME2 analysis [37]. The quoted values denote the median and corresponding 68% credible intervals.
This interpretation is further supported by the posterior distribution of shown in the fourth panel of Figure 7. In contrast to the pronounced bimodal structures of the individual and posteriors, the distribution of their product is single-peaked and considerably smoother. We obtain in the present analysis, compared with in the previous fixed-DDME2 calculation. The substantial overlap between the two distributions indicates that the bimodal behavior of and does not represent distinct physical branches, but instead reflects their non-identifiability when treated as separate parameters. The combination therefore provides a more direct measure of the gravity-sector contribution and a more meaningful quantity for comparison between the two analyses.
The most significant difference between the two analyses remains in . In the fixed-DDME2 analysis, was tightly localized around , whereas the present joint inference gives with a substantially broader posterior. This broadening is consistent with an increased matter–gravity degeneracy when the dense matter EOS is allowed to vary simultaneously with the parameters. Since the two analyses differ not only in their EOS treatment but also in the nuclear and astrophysical information entering the likelihood, the comparison should be regarded as illustrative rather than as a controlled EOS-model test.
Conclusion
In this work, we have performed a joint Bayesian analysis of the dense matter equation of state and neutron star structure within the framework of symmetric teleparallel gravity. For the matter sector, we adopted a Taylor expansion EOS around the nuclear saturation density, treating the nuclear matter parameters as free quantities. For the gravity sector, we considered the linear model and constrained its parameters simultaneously with the EOS parameters. The Bayesian inference combines complementary information from nuclear saturation properties, chiral effective field theory calculations of pure neutron matter, empirical and heavy-ion collision constraints, the GW170817 tidal deformability posterior, and the available NICER mass–radius measurements. In this way, the dense matter and modified-gravity parameters are inferred consistently within a common framework.
The resulting NS properties are compatible with the astrophysical constraints included in the analysis. For a canonical 1.4 star, we obtain km and , while the maximum NS mass is constrained to . The corresponding central properties are and . These results show that the posterior EOSs comfortably support neutron stars above the mass scale. The inferred correlations further reveal a strong relation between and , whereas remains only weakly correlated with the canonical star observables. Within the gravity sector, and exhibit a pronounced correlation, while the direct correlations of the modified gravity parameters with the stellar observables remain generally weak. The present analysis therefore provides a more complete picture of the EOS–gravity interplay in neutron stars. Allowing the dense matter EOS to vary simultaneously with the parameters makes it possible to quantify how the observational constraints are distributed between the matter and gravity sectors. Canonical star observables such as and primarily probe the EOS at intermediate supranuclear densities, whereas remains more sensitive to the less constrained high-density sector. Our results show that reliable constraints on modified gravity from neutron stars require the matter and gravity sectors to be inferred simultaneously, rather than fixing one sector while constraining the other. This joint treatment constitutes the main advance of the present work and provides a systematic framework for confronting symmetric teleparallel gravity with nuclear and neutron star observations.
Acknowledgements
NP and KZ acknowledge support by the CUHK-Shenzhen University development fund under grant No. UDF01003041 and UDF03003041, and Shenzhen Peacock fund under No. 2023TC0179, Ministry of Science and Technology of China under Grant No. 2024YFA1611004, NSFC fund under No. 92570117. PKS acknowledges IUCAA, Pune, India for providing support through the visiting Associateship program.
References
Appendix: Comparison with the Previous Fixed-DDME2 Analysis
For completeness, we compare the stellar predictions obtained in the present joint Taylor expansion EOS analysis with those from our previous fixed-DDME2 calculation [37] for the linear model. Since the two analyses differ in both the EOS treatment and the information entering the likelihood, the comparison is intended to illustrate the resulting changes in the posterior predictions rather than to provide a controlled EOS model comparison.
Figure 8 compares the 95% credible regions of the mass–radius and mass–tidal-deformability relations. The present Taylor expansion EOS analysis yields visibly narrower posterior bands over most of the relevant stellar mass range, while remaining broadly consistent with the corresponding fixed-DDME2 predictions. The reduced spread is consistent with the additional nuclear and multimessenger constraints restricting the allowed dense matter parameter space in the joint inference. The corresponding one-dimensional marginalized posterior distributions of , , and are shown in Figure 9. The median values of and remain close to those obtained in the fixed-DDME2 analysis, whereas their credible intervals are noticeably reduced in the present calculation. The maximum-mass distribution also becomes narrower and shifts toward a slightly lower median value. These results are consistent with the additional EOS constraints reducing the range of stellar configurations allowed by the joint analysis, including the high-density sector relevant to .

Figure 9. Similarly to Figure 7, but showing the one-dimensional marginalized posterior distributions of the neutron-star properties , , and . The quoted values denote the medians and corresponding 68% credible intervals.

Figure 8. Similarly to Figure 7, but showing the 95% credible regions of the neutron-star mass–radius relation (left panel) and mass–tidal-deformability relation (right panel).
References
- [1]C. M. Will, Living Rev. Rel. 17, 4 (2014), arXiv:1403.7377 [gr-qc] .DOI
- [2]A. G. Riess et al., Astron. J. 116, 1009 (1998), arXiv:astro-ph/9805201 .DOI
- [3]P. J. E. Peebles and B. Ratra, Rev. Mod. Phys. 75, 559 (2003), arXiv:astro-ph/0207347 .DOI
- [4]T. P. Sotiriou and V. Faraoni, Rev. Mod. Phys. 82, 451 (2010), arXiv:0805.1726 [gr-qc] .DOI
- [5]J. Beltrán Jiménez, L. Heisenberg, and T. S. Koivisto, Universe 5, 173 (2019), arXiv:1903.06830 [hep-th] .arxiv.org/abs/1903.06830
- [6]R. Aldrovandi and J. G. Pereira, Teleparallel Gravity: An Introduction, Fundamental Theories of Physics, Vol. 173 (Springer, Dordrecht, 2013).
- [7]S. Bahamonde et al., Rept. Prog. Phys. 86, 026901 (2023), arXiv:2106.13793 [gr-qc] .DOI
- [8]J. Beltrán Jiménez, L. Heisenberg, and T. Koivisto, Phys. Rev. D 98, 044048 (2018), arXiv:1710.03116 [gr-qc]DOI
- [9]S. Capozziello, V. De Falco, and C. Ferrara, Eur. Phys. J. C 82, 865 (2022), arXiv:2208.03011 [gr-qc] .DOI
- [10]L. Heisenberg, Phys. Rept. 1066, 1 (2024), arXiv:2309.15958 [gr-qc] .arxiv.org/abs/2309.15958
- [11]J. Beltrán Jiménez, L. Heisenberg, T. S. Koivisto, and S. Pekar, Phys. Rev. D 101, 103507 (2020), arXiv:1906.10027 [gr-qc] .DOI
- [12]F. K. Anagnostopoulos, V. Gakis, E. N. Saridakis, and S. Basilakos, Eur. Phys. J. C 83, 58 (2023), arXiv:2205.11445 [gr-qc] .DOI
- [13]R.-H. Lin and X.-H. Zhai, Phys. Rev. D 103, 124001 (2021), arXiv:2105.01484 [gr-qc] .DOI
- [14]W. Wang, H. Chen, and T. Katsuragawa, Phys. Rev. D 105, 024060 (2022), arXiv:2110.13565 [gr-qc] .arxiv.org/abs/2110.13565
- [15]D. Zhao, Eur. Phys. J. C 82, 303 (2022), arXiv:2104.02483 [gr-qc] .DOI
- [16]B. P. Abbott et al. (LIGO Scientific, Virgo), Phys. Rev. Lett. 119, 161101 (2017), arXiv:1710.05832 [gr-qc] .
- [17]F. Acernese et al. (Virgo), Class. Quant. Grav. 32, 024001 (2015), arXiv:1408.3978 [gr-qc] .DOI
- [18]B. P. Abbott et al. (LIGO Scientific, Virgo, Fermi GBM, INTEGRAL, IceCube, and others), Astrophys. J. Lett. 848, L12 (2017), arXiv:1710.05833 [astro-ph.HE] .
- [19]T. Hinderer, Astrophys. J. 677, 1216 (2008), arXiv:0711.2420 [astro-ph] .DOI
- [20]M. Oertel, M. Hempel, T. Klähn, and S. Typel, Rev. Mod. Phys. 89, 015007 (2017), arXiv:1610.03361 [astro-ph.HE] .DOI
- [21]K. Chatziioannou, Gen. Rel. Grav. 52, 109 (2020), arXiv:2006.03168 [gr-qc] .
- [22]T. E. Riley et al., Astrophys. J. Lett. 887, L21 (2019), arXiv:1912.05702 [astro-ph.HE] .DOI
- [23]M. C. Miller et al., Astrophys. J. Lett. 887, L24 (2019), arXiv:1912.05705 [astro-ph.HE] .DOI
- [24]T. E. Riley et al., Astrophys. J. Lett. 918, L27 (2021), arXiv:2105.06980 [astro-ph.HE] .DOI
- [25]D. Choudhury et al., Astrophys. J. Lett. 971, L20 (2024), arXiv:2407.06789 [astro-ph.HE] .DOI
- [26]L. Mauviard et al., Astrophys. J. 995, 60 (2025), arXiv:2506.14883 [astro-ph.HE] .DOI
- [27]B. Biswas, P. Char, R. Nandi, and S. Bose, Phys. Rev. D 103, 103015 (2021), arXiv:2008.01582 [astro-ph.HE] .DOI
- [28]B. Biswas, E. Smyrniotis, I. Liodis, and N. Stergioulas, Phys. Rev. D 109, 064048 (2024), arXiv:2309.05420 [gr-qc] .arxiv.org/abs/2309.05420
- [29]J. Margueron, R. Hoffmann Casali, and F. Gulminelli, Phys. Rev. C 97, 025805 (2018), arXiv:1708.06894 [nucl-th] .DOI
- [30]K. Hebeler, J. M. Lattimer, C. J. Pethick, and A. Schwenk, Astrophys. J. 773, 11 (2013), arXiv:1303.4662 [astro-ph.SR] .DOI
- [31]C. Drischler, K. Hebeler, and A. Schwenk, Phys. Rev. Lett. 122, 042501 (2019), arXiv:1710.08220 [nucl-th] .DOI
- [32]P. Danielewicz, R. Lacey, and W. G. Lynch, Science 298, 1592 (2002), arXiv:nucl-th/0208016 .DOI
- [33]M. B. Tsang, Y. Zhang, P. Danielewicz, M. Famiano, Z. Li, W. G. Lynch, and A. W. Steiner, Phys. Rev. Lett. 102, 122701 (2009), arXiv:0811.3107 [nucl-ex] .
- [34]N. Rutherford et al., Astrophys. J. Lett. 971, L19 (2024), arXiv:2407.06790 [astro-ph.HE] .DOI
- [35]L. Brandes and W. Weise, Phys. Rev. D 111, 034005 (2025), arXiv:2412.05923 [nucl-th] .DOI
- [36]S. M. A. Imam and N. K. Patra, Phys. Rev. D 112, 103018 (2025), arXiv:2509.07109 [nucl-th] .DOI
- [37]S. Pradhan, N. K. Patra, K. Zhou, and P. K. Sahoo, Eur. Phys. J. C 86, 684 (2026), arXiv:2601.16227 [gr-qc] .DOI
- [38]K. Nobleson, S. Banik, and T. Malik, Phys. Rev. D 107, 124045 (2023), arXiv:2306.01054 [gr-qc] .DOI
- [39]R. V. Lobato, G. A. Carvalho, and C. A. Bertulani, Eur. Phys. J. C 81, 1013 (2021), arXiv:2106.01841 [gr-qc] .DOI
- [40]R. Lobato, O. Lourenço, P. H. R. S. Moraes, C. H. Lenzi, M. de Avellar, W. de Paula, M. Dutra, and M. Malheiro, JCAP 12, 039 (2020), arXiv:2009.04696 [astro-ph.HE] .DOI
- [41]L.-W. Chen, C. M. Ko, and B.-A. Li, Phys. Rev. C 72, 064309 (2005), arXiv:nucl-th/0509009 .DOI
- [42]L.-W. Chen, B.-J. Cai, C. M. Ko, B.-A. Li, C. Shen, and J. Xu, Phys. Rev. C 80, 014322 (2009), arXiv:0905.4323 [nucl-th] .DOI
- [43]W. G. Newton, J. Hooker, M. Gearheart, K. Murphy, D.-H. Wen, F. J. Fattoyev, and B.-A. Li, Eur. Phys. J. A 50, 41 (2014), arXiv:1506.02207 [nucl-th] .DOI
- [44]J. Margueron, R. Hoffmann Casali, and F. Gulminelli, Phys. Rev. C 97, 025806 (2018), arXiv:1708.06895 [nucl-th] .DOI
- [45]J. Margueron and F. Gulminelli, Phys. Rev. C 99, 025806 (2019), arXiv:1807.01729 [nucl-th] .DOI
- [46]S. M. A. Imam, N. K. Patra, C. Mondal, T. Malik, and B. K. Agrawal, Phys. Rev. C 105, 015806 (2022), arXiv:2110.15776 [nucl-th] .DOI
- [47]N. K. Patra, S. M. A. Imam, B. K. Agrawal, A. Mukherjee, and T. Malik, Phys. Rev. D 106, 043024 (2022), arXiv:2203.08521 [nucl-th] .DOI
- [48]G. Baym, C. Pethick, and P. Sutherland, Astrophys. J. 170, 299 (1971).
- [49]R. C. Tolman, Phys. Rev. 55, 364 (1939).
- [50]J. R. Oppenheimer and G. M. Volkoff, Phys. Rev. 55, 374 (1939).
- [51]L. Heisenberg and C. Pastor-Marcos, “Neutron stars in f(Q) gravity,” (2025), arXiv:2512.03037 [gr-qc] .DOI
- [52]I. Ayuso, M. Bouhmadi-López, C.-Y. Chen, X. Y. Chew, K. Dialektopoulos, and Y. C. Ong, JCAP 11, 068 (2025), arXiv:2506.03506 [gr-qc] .DOI
- [53]K. Heebler, J. M. Lattimer, C. J. Pethick, and A. Schwenk, Astrophys. J. 773, 11 (2013), arXiv:1303.4662 [astro-ph.SR] .
- [54]C. Drischler, J. W. Holt, and C. Wellenhofer, Ann. Rev. Nucl. Part. Sci. 71, 403 (2021), arXiv:2101.01709 [nucl-th] .DOI
- [55]C. Drischler, R. J. Furnstahl, J. A. Melendez, and D. R. Phillips, Phys. Rev. Lett. 125, 202702 (2020), arXiv:2004.07232 [nucl-th] .DOI
- [56]C. Y. Tsang, M. B. Tsang, W. G. Lynch, R. Kumar, and C. J. Horowitz, Nature Astron. 8, 328 (2024), arXiv:2310.11588 [nucl-th] .DOI
- [57]S. M. A. Imam, T. Malik, C. Providência, and B. K. Agrawal, Phys. Rev. D 109, 103025 (2024), arXiv:2401.06018 [nucl-th] .DOI
- [58]S. M. A. Imam, P. Saxena, T. Malik, N. K. Patra, and B. K. Agrawal, Phys. Rev. D 110, 103042 (2024), arXiv:2407.08553 [nucl-th] .arxiv.org/abs/2407.08553
- [59]B. P. Abbott et al. (LIGO Scientific, Virgo), Phys. Rev. Lett. 119, 161101 (2017), arXiv:1710.05832 [gr-qc] .
- [60]T. E. Riley et al., Astrophys. J. Lett. 887, L21 (2019).
- [61]M. C. Miller et al., Astrophys. J. Lett. 887, L24 (2019).
- [62]T. E. Riley et al., Astrophys. J. Lett. 918, L27 (2021), arXiv:2105.06980 [astro-ph.HE] .DOI
- [63]M. C. Miller et al., Astrophys. J. Lett. 918, L28 (2021), arXiv:2105.06979 [astro-ph.HE] .DOI
- [64]D. Choudhury et al., Astrophys. J. Lett. 971, L20 (2024).
- [65]L. Mauviard et al., Astrophys. J. 995, 60 (2025).