betaFeedback

WorksPhysics and Astronomy

On the Possibility of a Strong First-Order Phase Transition in Neutron Stars

Abstract

Whether cold dense QCD matter undergoes a strong first-order phase transition remains an open question. In nature, neutron stars provide the most direct probe of cold dense QCD matter. Theoretically, chiral effective field theory constrains the equation of state of dense matter near nuclear saturation density, while perturbative QCD calculations constrain it at densities well beyond stable neutron-star interiors. We perform Bayesian inference with non-parametric Gaussian-process equation of state for ββ-equilibrated neutron-star matter under the assumption with and without a strong first-order phase transition, using the tidal deformability from GW170817, the NICER mass–radius measurements of PSR~J0740++6620, PSR~J0030++0451, PSR~J0437−-4715, PSR~J0614−-3329, chiral effective field theory, and perturbative QCD. Within this Bayesian inference framework, the model comparison mildly to moderately favors a strong first-order phase transition, with its onset most likely lying above the central density of the most massive neutron star. Such an onset reconciles the stiffness required to support massive neutron stars with the softening favored by perturbative QCD from asymptotically high density.

Introduction

The phase structure of strong interaction matter at several times nuclear saturation density, n0≡0.16 fm−3n_0 \equiv0.16\,\mathrm{fm}^{-3}, remains a central open problem in nuclear physics and astrophysics. At low baryon chemical potential and high temperature, lattice quantum chromodynamics (QCD) and ultrarelativistic heavy-ion experiments have established that the transition from hadronic matter to the quark-gluon plasma is an analytic crossover [11, 21, 10, 19, 16]. At low temperature and high baryon density, however, first-principles lattice calculations are obstructed by the sign problem [6, 43]. Whether cold dense QCD matter contains a strong first-order phase transition (FOPT), characterized here by a vanishing sound speed over a finite density interval as in a sharp Maxwell construction

Corresponding author: Lie-Wen Chen

link text

tion [34], is therefore still unknown from first principles.

Neutron stars provide the most direct observational access to cold dense matter. Recent multi-messenger measurements have substantially narrowed the uncertainties of equation of state (EOS) of cold dense β\beta-equilibrated matter. In particular, the gravitational wave signal GW170817 from the binary neutron-star merger constrains the tidal deformability of canonical-mass neutron stars [1] (Abbott et al. 2019), while the Neutron Star Interior Composition Explorer (NICER) provides simultaneous mass–radius measurements for several pulsars through pulse-profile modeling. The NICER measurement of the high-mass pulsar PSR J0740+6620 [83] (Miller et al. 2021; Riley et al. 2021) yields a mass M≃2.08 M⊙M \simeq2.08\,M_\odot that places a stringent lower bound on the neutron-star maximum mass, while its measured radius constrains the EOS at high density. The lower-mass NICER sources PSR J0030+0451 [68, 79] (Vinciguerra et al. 2024), PSR J0437−4715 (Choudhury et al. 2024a; Reardon et al. 2024), and PSR J0614−3329 ([66]) further constrain the neutron-star radii in the canonical-mass regime. Together, current observations probe the EOS of neutron-star matter over baryon densities of roughly 33–8n08n_0 reached inside stable neutron stars.

Theoretical calculations constrain the EOS in complementary density regimes. At low density, chiral effective field theory (ChEFT) provides controlled calculations up to ∼1\sim1–2n02n_0, with quantified many-body and truncation uncertainties ([95, 44, 64, 27, 26, 51]). At very high density, perturbative QCD (pQCD) provides a weak-coupling calculation of the EOS once the quark chemical potential is sufficiently large ([31, 59, 60, 39, 38, 37]). Although pQCD thermodynamics becomes quantitatively reliable only at extremely high densities of order 40n040n_0, causality and thermodynamic stability allow the high-density constraint to be propagated downward ([55]). Recent work has shown that the pQCD speed of sound remains well converged down to about 25n025n_0 ([37]), extending the validity range of pQCD toward lower densities. Nevertheless, in the broad density range between the ChEFT and pQCD regimes, it is still unclear whether the EOS varies smoothly (no strong first-order phase transition, NPT) or contains an FOPT.

These observational and theoretical inputs have sparked extensive studies of the FOPT in cold dense QCD matter. Most analyses ([5, 7, 12, 87, 70, 42, 24, 74, 17, 9, 94, 92, 99, 35, 20, 91, 29, 58, 8, 25, 104, 82, 13, 40, 100, 63, 81, 41, 28] (Ji et al. 2025; Li et al. 2025)), however, do not directly compare the FOPT and NPT hypotheses through Bayesian model selection. Such a comparison is carried out in only a few studies ([75, 56, 48, 93]), but the FOPT onset density they consider is generally restricted to densities reached in neutron-star interiors. It is therefore worthwhile to test the FOPT hypothesis over a broader density range that spans the full ChEFT–pQCD window.

In this work, we perform Bayesian inference using a non-parametric Gaussian-process (GP) EOS for cold β\beta-equilibrated neutron-star matter. The inference combines the tidal deformability of GW170817, NICER mass–radius measurements of PSR J0740+6620, PSR J0030+0451, PSR J0437−4715, and PSR J0614−3329, ChEFT below 1.5n01.5n_0, and pQCD constraints. The GP EOS is terminated at nL=25n0n_L=25n_0 to span the full ChEFT–pQCD window, with a lower choice nL=12n0n_L=12n_0 examined for comparison. Comparing the NPT hypothesis with the FOPT hypothesis, we find which cs2=0c_s^2=0 over a finite density interval, we find mild to moderate evidence in favor of an FOPT over NPT, with the FOPT onset density most likely located above the central density of the maximum-mass neutron star.

This paper is organized as follows. § 2 describes the Gaussian-process EOS construction and the Bayesian framework combining GW170817, NICER, ChEFT, and pQCD constraints. § 3 presents the inferred FOPT properties and their implications for the EOS, stellar observables, and the location of the transition relative to stable neutron-star interiors. § 4 summarizes the main conclusions.

Methods

Physics-agnostic EOS

To construct a physics-agnostic and non-parametric EOS of cold β\beta-equilibrated neutron-star matter, we model its squared sound speed cs2(n)c_s^2(n) at baryon number density nn with GP regression ([77]). Causality and thermodynamic stability bound cs2c_s^2 to [0,1][0,1], so we place the GP prior on the auxiliary variable φ(n)≡−ln⁡[1/cs2(n)−1]\varphi(n)\equiv-\ln[1/c_s^2(n)-1] rather than on cs2c_s^2 itself,

φ(n)∼GP(φˉ,K(n,n′)),φˉ=−ln⁡[1/cˉs2−1],(1)\varphi(n)\sim\mathcal{GP}\left(\bar{\varphi},K(n,n')\right),\qquad\bar{\varphi}=-\ln\left[1/\bar{c}_s^2-1\right], \tag*{(1)}

with squared-exponential kernel K(n,n′)=ηexp⁡[−(n−n′)2/(2ℓ2)]K(n,n')=\eta\exp[-(n-n')^2/(2\ell^2)]. The hyperprior distributions are ([36])

ℓ∼N(1.0n0,(0.25n0)2),\ell\sim\mathcal{N}\left(1.0n_0,(0.25n_0)^2\right),
η∼N(1.25,0.22),\eta\sim\mathcal{N}\left(1.25,0.2^2\right),
cˉs2∼N(0.5,0.252).(2)\bar{c}_s^2\sim\mathcal{N}\left(0.5,0.25^2\right). \tag*{(2)}

The GP variable φ(n)\varphi(n) is predicted over densities from a crust-matching density ncc=0.5n0n_{\rm cc}=0.5n_0 up to a termination density nLn_L. Since pQCD constrains a nonzero cs2c_s^2 down to 25n025n_0 (e.g., Fig. 10 of Komoltsev et al. [56]), we set nL=25n0n_L=25n_0 to span the full ChEFT–pQCD interval. We also examine nL=12n0n_L=12n_0 for comparison.

The GP of Eq. (1) is conditioned on theoretical calculations at a set of training densities $\mathbf{n}_t=\mathcal{T}_{\rm ChEFT}\cup TpQCD\mathcal{T}_{\mathrm{pQCD}}, comprising a low-density ChEFT subset TChEFT\mathcal{T}_{\mathrm{ChEFT}} and a high-density pQCD subset TpQCD\mathcal{T}_{\mathrm{pQCD}}. The ChEFT subset is a grid aligned with the next-to-next-to-next-to-leading-order (N3^{3}LO) ChEFT calculation of Keller et al. [51], spanning 0.58n0≤n≤1.5n00.58 n_{0} \le n \le1.5 n_{0}. At each nt∈TChEFTn_{t} \in\mathcal{T}_{\mathrm{ChEFT}}, the mean cs,t2c_{s,t}^{2} is the midpoint of the softest and stiffest ChEFT sound speeds, and σt\sigma_{t} is fixed so that the 90% interval of N(cs,t2,σt2)\mathcal{N}(c_{s,t}^{2},\sigma_{t}^{2}) matches the ChEFT band. The pQCD subset TpQCD={25n0}\mathcal{T}_{\mathrm{pQCD}} = \{25 n_{0}\} anchors the GP to the pQCD prediction at 25n025 n_{0}, where cs2(25n0)∼N(cs,t2,σt2)c_{s}^{2}(25 n_{0}) \sim\mathcal{N}(c_{s,t}^{2},\sigma_{t}^{2}). Following Komoltsev et al. [56], we take cs,t2=0.32 c2c_{s,t}^{2} = 0.32\,c^{2} as the mean and σt=0.05 c2\sigma_{t} = 0.05\,c^{2} as twice the standard deviation of the pQCD predictions at 25n025 n_{0} [52].

Each cs2(n)c_{s}^{2}(n) realization is drawn from the GP conditioned on the training data. The training inputs are mapped into the GP variable through ϕt=ϕ(cs,t2)\phi_{t} = \phi(c_{s,t}^{2}), while propagating σt\sigma_{t} through ϕ\phi gives the diagonal training covariance Σt=diag⁡{14[ϕ(cs,t2+σt)−ϕ(cs,t2−σt)]2}\boldsymbol{\Sigma}_{t} = \operatorname{diag}\left\{\frac{1}{4}\left[\phi(c_{s,t}^{2}+\sigma_{t})-\phi(c_{s,t}^{2}-\sigma_{t})\right]^{2}\right\}. Conditioning the GP prior on ϕt\phi_{t} and Σt\boldsymbol{\Sigma}_{t} then yields the predicted values ϕ⋆≡ϕ(n⋆)\phi_{\star} \equiv\phi(n_{\star}) at prediction densities n⋆∈[ncc,nL]\mathbf{n}_{\star} \in[n_{\mathrm{cc}},n_{L}], following a multivariate Gaussian,

p(ϕ⋆∣ϕt,Σt)=N(μ⋆,Σ⋆),(3)p(\phi_{\star}\mid\phi_{t},\boldsymbol{\Sigma}_{t})=\mathcal{N}(\boldsymbol{\mu}_{\star},\boldsymbol{\Sigma}_{\star}), \tag*{(3)}

with mean and covariance

μ⋆=ϕˉ+K(n⋆,nt)C−1(ϕt−ϕˉ),(4)\boldsymbol{\mu}_{\star}=\bar{\phi}+\mathbf{K}(\mathbf{n}_{\star},\mathbf{n}_{t})\mathbf{C}^{-1}(\phi_{t}-\bar{\phi}), \tag*{(4)}
Σ⋆=K(n⋆,n⋆)−K(n⋆,nt)C−1K(nt,n⋆),(5)\boldsymbol{\Sigma}_{\star}=\mathbf{K}(\mathbf{n}_{\star},\mathbf{n}_{\star})-\mathbf{K}(\mathbf{n}_{\star},\mathbf{n}_{t})\mathbf{C}^{-1}\mathbf{K}(\mathbf{n}_{t},\mathbf{n}_{\star}), \tag*{(5)}

where C≡K(nt,nt)+Σt\mathbf{C} \equiv\mathbf{K}(\mathbf{n}_{t},\mathbf{n}_{t})+\boldsymbol{\Sigma}_{t}. Sampling ϕ⋆\phi_{\star} from p(ϕ⋆∣ϕt,Σt)p(\phi_{\star}\mid\phi_{t},\boldsymbol{\Sigma}_{t}) and inverting the map ϕ\phi yields cs2(n)c_{s}^{2}(n).

The construction above corresponds to the NPT hypothesis, where the GP yields a smooth cs2(n)c_{s}^{2}(n) throughout [ncc,nL][n_{\mathrm{cc}},n_{L}]. For the FOPT, we adopt a uniform-density prior for the phase-transition location by sampling two densities uniformly over [1.5n0,nL][1.5 n_{0},n_{L}] and labeling the smaller and larger values as the phase-transition onset density nSn_{S} and end density nEn_{E}, respectively. In the Maxwell construction, the sound speed vanishes between them,

cs2(n)=0,nS≤n≤nE.(6)c_{s}^{2}(n)=0,\qquad n_{S}\le n\le n_{E}. \tag*{(6)}

We then draw two independent GP realizations over [ncc,nL][n_{\mathrm{cc}},n_{L}] and take the first on [ncc,nS][n_{\mathrm{cc}},n_{S}] and the second on [nE,nL][n_{E},n_{L}], leaving the sound speed on the two sides of the transition uncorrelated.

Once cs2(n)c_{s}^{2}(n) is specified for n≥nccn\ge n_{\mathrm{cc}} under either hypothesis, the EOS follows from the thermodynamic relations

μ(n)=μ(ncc)exp⁡[∫nccndn′ cs2(n′)n′],(7)\mu(n)=\mu(n_{\mathrm{cc}})\exp\left[\int_{n_{\mathrm{cc}}}^{n}\mathrm{d}n'\,\frac{c_{s}^{2}(n')}{n'}\right], \tag*{(7)}
ϵ(n)=ϵ(ncc)+∫nccndn′ μ(n′),(8)\epsilon(n)=\epsilon(n_{\mathrm{cc}})+\int_{n_{\mathrm{cc}}}^{n}\mathrm{d}n'\,\mu(n'), \tag*{(8)}
p(n)=−ϵ(n)+μ(n)n,(9)p(n)=-\epsilon(n)+\mu(n)n, \tag*{(9)}

where μ\mu, ϵ\epsilon, and pp denote the baryon chemical potential, energy density, and pressure, respectively. For n≤nccn\le n_{\mathrm{cc}}, we adopt the crust EOS [15, 72], which fixes the boundary values μ(ncc)\mu(n_{\mathrm{cc}}) and ϵ(ncc)\epsilon(n_{\mathrm{cc}}) in Eqs. (7) and (8).

We solve the Tolman–Oppenheimer–Volkoff equations [97, 73] for each EOS sample to obtain the neutron-star mass–radius relation and the central baryon density ncn_{c} of the most massive configuration. Within the FOPT hypothesis, we further distinguish whether the transition occurs inside stable neutron stars. Samples with nS<ncn_{S}<n_{c} are labeled FOPT-in, while samples with nS≥ncn_{S}\ge n_{c} are labeled FOPT-out. The dimensionless tidal deformability Λ\Lambda is obtained from the ℓ=2\ell=2 tidal Love number k2k_{2} [47]. Under the FOPT hypothesis, the energy-density discontinuity Δϵ\Delta\epsilon at the transition radius rdr_{d} shifts the metric perturbation variable y≡rH′/Hy\equiv rH'/H by Δy=−4πrd3Δϵ/[m(rd)+4πrd3p(rd)]\Delta y=-4\pi r_{d}^{3}\Delta\epsilon/[m(r_{d})+4\pi r_{d}^{3}p(r_{d})] [76], which we incorporate when integrating k2k_{2} across the transition.

An FOPT EOS may give rise to two stable branches in the mass–radius relation, allowing for twin-star configurations. To detect this behavior, we solve the TOV equations on a grid of compact-star central pressures pcp_{c} spaced log-uniformly from 0.5 MeV fm−30.5\,\mathrm{MeV}\,\mathrm{fm}^{-3} up to the pressure at the termination density nLn_{L} (using 150 points for nL=25n0n_{L}=25 n_{0} and 100 points for nL=12n0n_{L}=12 n_{0}), and identify a second stable branch whenever the mass M(pc)M(p_{c}) rises again over at least three consecutive solutions after passing through a local minimum. It should be emphasized that the Seidov condition [86] Δϵcrit/ϵt=12+32pt/ϵt\Delta\epsilon_{\mathrm{crit}}/\epsilon_{t}=\frac{1}{2}+\frac{3}{2}p_{t}/\epsilon_{t} (with ϵt\epsilon_{t} and ptp_{t} the energy density and pressure at the transition), often invoked for an FOPT, does not directly determine the existence of a twin star. In fact, stable second branches can arise for both Δϵ<Δϵcrit\Delta\epsilon<\Delta\epsilon_{\mathrm{crit}} and Δϵ>Δϵcrit\Delta\epsilon>\Delta\epsilon_{\mathrm{crit}}, as illustrated by Figure 2(b) and Figure 2(d), respectively, of Alford et al. [5]. We therefore classify twin stars directly from the stability of the computed sequence in the following.

Image not extracted

Figure 2. Joint posterior distributions of mass–radius (top) and tidal deformability–mass (bottom) for neutron stars under the FOPT (blue) and NPT (red) hypotheses, shown as 68% and 95% credible regions. The left (right) column is for nL=25n0n_L = 25 n_0 (12n012 n_0). Astrophysical constraints—NICER joint posteriors for PSR J0740+6620 ([83]), PSR J0030+0451 ((Vinciguerra et al. 2024)), PSR J0437−4715 ([22]), and PSR J0614−3329 ([66]), together with the GW170817 posterior ((Abbott et al. 2018))—are overlaid for reference.

Bayesian inference

We combine the theoretical and observational information in a hierarchical Bayesian framework. For data d⃗\vec{d} and hypothesis H\mathcal{H}, the posterior for the EOS parameters θ\theta is [96, 45]

p(θ∣d⃗,H)=∏iL(di∣θ,H) π(θ∣H)ZH(d⃗),(10)p(\theta\mid\vec{d}, \mathcal{H}) = \frac{\prod_i \mathcal{L}(d_i \mid\theta, \mathcal{H})\,\pi(\theta\mid\mathcal{H})}{\mathcal{Z}_{\mathcal{H}}(\vec{d})}, \tag*{(10)}

where ii runs over independent constraints and π(θ∣H)\pi(\theta\mid\mathcal{H}) denotes the prior under hypothesis H\mathcal{H}. The evidence is

ZH(d⃗)≡∫∏iL(di∣θ,H) π(θ∣H) dθ,(11)\mathcal{Z}_{\mathcal{H}}(\vec{d}) \equiv\int\prod_i \mathcal{L}(d_i \mid\theta, \mathcal{H})\,\pi(\theta\mid\mathcal{H})\,\mathrm{d}\theta, \tag*{(11)}

and it quantifies the support that the data give to the hypothesis. We compare two hypotheses through the Bayes factor [49, 61]

B21=Z1(d⃗)/Z2(d⃗).(12)B_2^1 = \mathcal{Z}_1(\vec{d})/\mathcal{Z}_2(\vec{d}). \tag*{(12)}

On the Jeffreys scale [49, 98, 61], B21B_2^1 in the ranges [1, 3], [3, 10], [10, 30], [30, 100], and [100, ∞\infty) corresponds to anecdotal, moderate, strong, very strong, and extreme evidence, respectively, in favor of hypothesis 1.

For each NICER mass–radius measurement, the likelihood is obtained by marginalizing a kernel density estimate, with the bandwidth set by Scott’s rule [85], of the published mass–radius posterior over the neutron-star mass along the EOS-predicted curve R(m,θ)R(m,\theta). The analysis includes PSR J0740+6620 [83 b], PSR J0030+0451 [101] (Vinciguerra et al. 2024), PSR J0437−4715 [22 b], and PSR J0614−3329 [66 b]. The published posteriors for PSR J0740+6620, PSR J0437−4715, and PSR J0614−3329 already include the corresponding radio-timing mass information, whereas PSR J0030+0451 has no independent radio-timing mass prior. For the tidal deformability from

GW170817 [1] (Abbott et al. 2019), we adopt the nuisance-marginalized likelihood of Hernandez Vivanco et al. [45] under the low-spin prior and marginalize over the component masses along the EOS-predicted curve Λ(m,θ)\Lambda(m,\theta). Throughout the Bayesian analysis, we adopt a flat prior on the compact-star mass mm.

The above astrophysical likelihoods constrain the EOS only within stable neutron stars (n≲5n \lesssim5–8n08n_0). To extend the constraint to higher densities, we adopt the marginalized pQCD likelihood of Komoltsev et al. [56, 54]. The likelihood is built from an auxiliary Gaussian process GPaux\mathrm{GP}_{\mathrm{aux}}, anchored to the pQCD thermodynamic quantities at npQCD≃10n0n_{\mathrm{pQCD}} \simeq10n_0 [38] and conditioned on the pQCD-predicted cs2(n)c_s^2(n) over 25n0≤n≤40n025n_0 \leq n \leq40n_0, where perturbative corrections remain small [37]. For any nm≤25n0n_m \leq25n_0, marginalizing GPaux\mathrm{GP}_{\mathrm{aux}} over [nm,npQCD][n_m,n_{\mathrm{pQCD}}] yields a joint distribution of (p,ϵ)(p,\epsilon) at nmn_m, which defines the marginalized pQCD likelihood at nmn_m. We evaluate this likelihood directly at the EOS termination density nLn_L, i.e., nm=nLn_m=n_L, thereby constraining the entire EOS below nLn_L. Two caveats are in order. First, the pQCD likelihood constrains (p,ϵ)(p,\epsilon) at nLn_L, reflecting the integral of cs2c_s^2 up to nLn_L rather than cs2(nL)c_s^2(n_L) itself. Second, GPaux\mathrm{GP}_{\mathrm{aux}} is smooth by construction and therefore does not model an FOPT between nLn_L and npQCDn_{\mathrm{pQCD}}. The likelihood terms, together with their references and data, are summarized in Table 1. In summary, the likelihood terms listed in Table 1, together with the CEFT sound-speed constraint at the low-density training set TCEFT\mathcal{T}_{\mathrm{CEFT}} and the pQCD sound-speed anchor at the training density TpQCD={25n0}\mathcal{T}_{\mathrm{pQCD}}=\{25n_0\}, comprise our default dataset d⃗def\vec{d}_{\mathrm{def}}.

LikelihoodSource
PSR J0740+6620Ref. [83 b]; ST-U hot-spot model.
PSR J0030+0451Ref. [101] (Vinciguerra et al. 2024); ST+PDT hot-spot model.
PSR J0437−4715Ref. [22 b]; CST+PDT hot-spot model.
PSR J0614−3329Ref. [66 b]; ST+PDT hot-spot model.
GW170817Ref. [1]; [45]; low-spin prior.
Marginalized pQCDRef. [56, 54].

Table 1. Likelihood terms used in the Bayesian inference. For each term, the second column lists the reference and data from which the likelihood is built, together with the hot-spot model (for the NICER mass–radius measurements) or the spin prior (for GW170817).

We evaluate posteriors and evidences by importance sampling. For each hypothesis H\mathcal{H} and each value of nLn_L, we draw N=8×106N=8\times10^6 samples {θk}\{\theta_k\} from the prior and assign each a weight wk=∏iL(di∣θk,H)w_k=\prod_i\mathcal{L}(d_i\mid\theta_k,\mathcal{H}). The evidence then follows as ZH(d⃗)≃N−1∑kwk\mathcal{Z}_{\mathcal{H}}(\vec{d})\simeq N^{-1}\sum_k w_k, and the effective sample size [57], Neff=N_{\mathrm{eff}}= (∑kwk)2/∑kwk2(\sum_{k} w_{k})^{2}/\sum_{k} w_{k}^{2}, exceeds 1×1041 \times10^{4} in all runs, ensuring convergence of the posterior estimates. Equally-weighted posterior samples are then obtained by resampling with replacement according to wkw_{k}, from which posterior distributions and Monte Carlo uncertainty are computed.

Results and Discussion

For nL=25n0n_{\mathrm{L}} = 25 n_{0} with the default dataset d⃗def\vec{d}_{\mathrm{def}}, we find BNPTFOPT=3.20±0.04B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 3.20 \pm0.04 (here and throughout, the quoted uncertainties associated with Bayes factors are Monte Carlo standard errors), corresponding to moderate evidence on the Jeffreys scale for the FOPT hypothesis relative to the NPT hypothesis. The prior and posterior distributions of the FOPT onset density nSn_{\mathrm{S}}, end density nEn_{\mathrm{E}}, and the central density ncn_{\mathrm{c}} of the maximum-mass neutron star are shown in Figure 1, where blue and orange samples correspond to the FOPT-in (nS<ncn_{\mathrm{S}} < n_{\mathrm{c}}) and FOPT-out (nS≥ncn_{\mathrm{S}} \ge n_{\mathrm{c}}), respectively. Within the FOPT hypothesis, the data further favor FOPT-out over FOPT-in, with Binout=4.01±0.21B_{\mathrm{in}}^{\mathrm{out}} = 4.01 \pm0.21. Together, for nL=25n0n_{\mathrm{L}} = 25 n_{0}, the data support an FOPT whose onset most likely lies above the central density of the maximum-mass neutron star. Placing the FOPT onset above ncn_{\mathrm{c}} agrees with earlier physics-agnostic inferences that incorporate heavy pulsars and find no sharp sound-speed drop within massive neutron-star cores ([88, 20]). This agreement reflects a common physical tension: supporting massive neutron stars demands a stiff EOS in their interiors, while a strong FOPT would soften the EOS.

Image not extracted

Figure 1. Corner plots of the posterior distributions of the FOPT onset density nSn_{\mathrm{S}}, end density nEn_{\mathrm{E}}, and central density ncn_{\mathrm{c}} of the most massive neutron star for the two choices of the termination density nLn_{\mathrm{L}}: nL=25n0n_{\mathrm{L}} = 25 n_{0} (left panel) and nL=12n0n_{\mathrm{L}} = 12 n_{0} (right panel). In each panel, the diagonal subpanels show the priors and one-dimensional marginalized posteriors, with the median and 68% credible interval reported in each panel title; the Bayes factors BNPTFOPTB_{\mathrm{NPT}}^{\mathrm{FOPT}} and BinoutB_{\mathrm{in}}^{\mathrm{out}} are also annotated. The off-diagonal panels display equally-weighted posterior samples, with blue (nS<ncn_{\mathrm{S}} < n_{\mathrm{c}}) and orange (nS≥ncn_{\mathrm{S}} \ge n_{\mathrm{c}}) marking FOPT onsets below and above ncn_{\mathrm{c}}, respectively.

The default dataset d⃗def\vec{d}_{\mathrm{def}} contains two high-density pQCD inputs: the sound-speed anchor at TpQCD={25n0}T_{\mathrm{pQCD}} = \{25 n_{0}\} and the marginalized pQCD likelihood. To isolate their respective effects, each input is removed in turn at nL=25n0n_{\mathrm{L}} = 25 n_{0}. Removing the sound-speed anchor from d⃗def\vec{d}_{\mathrm{def}} while retaining the marginalized pQCD likelihood, we obtain BNPTFOPT=3.18±0.04B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 3.18 \pm0.04 and Binout=4.11±0.29B_{\mathrm{in}}^{\mathrm{out}} = 4.11 \pm0.29. These values stay close to the results for d⃗def\vec{d}_{\mathrm{def}} (BNPTFOPT=3.20±0.04B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 3.20 \pm0.04 and Binout=4.01±0.21B_{\mathrm{in}}^{\mathrm{out}} = 4.01 \pm0.21), which indicates that the anchor at TpQCDT_{\mathrm{pQCD}} has little influence on our conclusions. Conversely, removing the marginalized pQCD likelihood from d⃗def\vec{d}_{\mathrm{def}} while retaining the anchor at TpQCDT_{\mathrm{pQCD}}, we find BNPTFOPT=0.79±0.01B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 0.79 \pm0.01 and Binout=2.53±0.14B_{\mathrm{in}}^{\mathrm{out}} = 2.53 \pm0.14, which differ substantially from the results for d⃗def\vec{d}_{\mathrm{def}}. The high-density pQCD constraint therefore acts predominantly through the marginalized likelihood rather than through the sound-speed anchor.

It is interesting to examine whether the astrophysical constraints and the pQCD constraint can each, on their own, drive the preference for an FOPT. Removing both the marginalized pQCD likelihood and the sound-speed anchor at TpQCDT_{\mathrm{pQCD}} from d⃗def\vec{d}_{\mathrm{def}}, we obtain BNPTFOPT=0.80±0.01\mathcal{B}_{\mathrm{NPT}}^{\mathrm{FOPT}} = 0.80 \pm0.01, slightly below unity, and Binout=2.15±0.11\mathcal{B}_{\mathrm{in}}^{\mathrm{out}} = 2.15 \pm0.11. Removing instead the astrophysical constraints, we find BNPTFOPT=1.04±0.01\mathcal{B}_{\mathrm{NPT}}^{\mathrm{FOPT}} = 1.04 \pm0.01 and Binout=1.72±0.02\mathcal{B}_{\mathrm{in}}^{\mathrm{out}} = 1.72 \pm0.02, both close to unity, which shows that the high-density pQCD constraint by itself yields no decisive preference. Neither the astrophysical data nor the pQCD constraint thus provides compelling evidence for an FOPT over NPT on its own. The support for an FOPT therefore emerges only from the interplay between the astrophysical constraints and the pQCD constraint.

In addition, to assess the sensitivity of the inferred preference for an FOPT to the prior specification, the analysis is repeated at nL=25n0n_L = 25 n_0 under three alternative choices: a log-uniform density prior and a chemical-potential prior for the FOPT location, as well as a broadened GP hyperprior. Under the log-uniform density prior, the densities nSn_S and nEn_E are sampled uniformly in ln⁡n\ln n over 1.5n0≤n≤nL1.5 n_0 \leq n \leq n_L and then ordered so that nS<nEn_S < n_E. Under the chemical-potential prior, a smooth GP EOS first defines μ(n)\mu(n), after which the transition chemical potential μt\mu_t is drawn uniformly over μ(1.5n0)≤μt≤μ(nL)\mu(1.5 n_0) \leq\mu_t \leq\mu(n_L) to determine nSn_S through μ(nS)=μt\mu(n_S) = \mu_t, and the latent heat Δε\Delta\varepsilon is drawn uniformly over 0≤Δε≤μt(nL−nS)0 \leq\Delta\varepsilon\leq\mu_t(n_L-n_S) to fix nE=nS+Δε/μt≤nLn_E = n_S+\Delta\varepsilon/\mu_t \leq n_L. For the log-uniform density and chemical-potential priors, the hyperprior specified in Eq. (2) is retained, whereas for the broadened GP hyperprior, the uniform-density prior for the FOPT location is retained and all three standard deviations are doubled while their means are held fixed. For the log-uniform density prior, the chemical-potential prior, and the broadened GP hyperprior, we find BNPTFOPT=2.24±0.03\mathcal{B}_{\mathrm{NPT}}^{\mathrm{FOPT}} = 2.24 \pm0.03, 2.65±0.042.65 \pm0.04, and 2.39±0.042.39 \pm0.04, respectively. The corresponding FOPT-out to FOPT-in Bayes factors are Binout=3.68±0.12\mathcal{B}_{\mathrm{in}}^{\mathrm{out}} = 3.68 \pm0.12, 3.32±0.223.32 \pm0.22, and 3.30±0.223.30 \pm0.22, respectively. Thus, changing the FOPT-location prior or broadening the GP hyperprior shifts the Bayes factors only at the order-unity level, and the results still favor both the FOPT hypothesis and an onset beyond the maximum-mass central density.

For comparison, we also perform the inference under d⃗def\vec{d}_{\mathrm{def}} with a more restricted extrapolation boundary nL=12n0n_L = 12 n_0. In this case, the support for the FOPT hypothesis drops to BNPTFOPT=2.84±0.03\mathcal{B}_{\mathrm{NPT}}^{\mathrm{FOPT}} = 2.84 \pm0.03, only anecdotal on the Jeffreys scale. The preference for FOPT-out also weakens, from Binout=4.01±0.21\mathcal{B}_{\mathrm{in}}^{\mathrm{out}} = 4.01 \pm0.21 at nL=25n0n_L = 25 n_0 to 1.88±0.071.88 \pm0.07 at nL=12n0n_L = 12 n_0. Because the marginalized pQCD likelihood is built from a smooth auxiliary GP above nLn_L, lowering nLn_L from 25n025 n_0 to 12n012 n_0 excludes any FOPT in the density range 12n0<n<25n012 n_0 < n < 25 n_0 and thereby reduces both Bayes factors. Even within this restricted window, nevertheless, the data still mildly favor an FOPT against NPT with the FOPT onset density lying above ncn_c.

The FOPT hypothesis shifts the maximum mass of a nonrotating neutron star MTOVM_{\mathrm{TOV}} modestly upward relative to the NPT. This shift is visible at the upper edge of the mass–radius contour in Figure 2(a), where MTOVNPT=2.07−0.08+0.10 M⊙M_{\mathrm{TOV}}^{\mathrm{NPT}} = 2.07^{+0.10}_{-0.08}\,M_\odot (68% credible interval, here and throughout unless stated otherwise) rises to MTOVFOPT=2.15−0.11+0.13 M⊙M_{\mathrm{TOV}}^{\mathrm{FOPT}} = 2.15^{+0.13}_{-0.11}\,M_\odot for nL=25n0n_L = 25 n_0. It should be noted that the marginalized pQCD likelihood at nLn_L disfavors stiff EOSs and is therefore in tension with the stiffness required to support a ∼2 M⊙\sim2\,M_\odot neutron star. An FOPT resolves this tension. The cs2=0c_s^2 = 0 plateau softens the EOS and relaxes the pQCD penalty, while the inferred onset lies predominantly above ncn_c, leaving the stellar interior stiff enough for MTOVM_{\mathrm{TOV}} to readily satisfy the PSR J0740+6620 mass constraint.

Although the FOPT hypothesis raises MTOVM_{\mathrm{TOV}}, the canonical-mass observables remain essentially unchanged. For nL=25n0n_{\mathrm{L}} = 25n_{0} we obtain R1.4FOPT=11.76−0.38+0.38 kmR_{1.4}^{\mathrm{FOPT}} = 11.76^{+0.38}_{-0.38}\,\mathrm{km} versus R1.4NPT=11.79−0.41+0.37 kmR_{1.4}^{\mathrm{NPT}} = 11.79^{+0.37}_{-0.41}\,\mathrm{km}, and Λ1.4FOPT=309−70+88\Lambda_{1.4}^{\mathrm{FOPT}} = 309^{+88}_{-70} versus Λ1.4NPT=317−75+86\Lambda_{1.4}^{\mathrm{NPT}} = 317^{+86}_{-75} [panel (c) of Fig. 2]—the two hypotheses are statistically indistinguishable at the 68% level. Earlier analyses that assume an FOPT at 22–3n03n_{0} soften the EOS at canonical-mass densities and drive R1.4R_{1.4} down to ∼10 km\sim10\,\mathrm{km} ([90, 89]). In our analysis, by contrast, nSn_{\mathrm{S}} is inferred rather than imposed, and the posterior favors an onset above the central density of the most massive neutron star, leaving the canonical-mass interior essentially unaffected.

Figure 2(b) and 2(d) show the mass–radius and tidal-deformability–mass joint posteriors for nL=12n0n_{\mathrm{L}} = 12n_{0}, which reproduce the qualitative conclusions established at nL=25n0n_{\mathrm{L}} = 25n_{0}. The canonical-mass observables remain statistically indistinguishable between the two hypotheses at the 68% level, with R1.4FOPT=11.79−0.39+0.39 kmR_{1.4}^{\mathrm{FOPT}} = 11.79^{+0.39}_{-0.39}\,\mathrm{km} versus R1.4NPT=11.75−0.39+0.39 kmR_{1.4}^{\mathrm{NPT}} = 11.75^{+0.39}_{-0.39}\,\mathrm{km} and Λ1.4FOPT=314−71+92\Lambda_{1.4}^{\mathrm{FOPT}} = 314^{+92}_{-71} versus Λ1.4NPT=308−70+89\Lambda_{1.4}^{\mathrm{NPT}} = 308^{+89}_{-70}. The modest FOPT-driven upward shift in the maximum mass also persists for nL=12n0n_{\mathrm{L}} = 12n_{0}, from MTOVNPT=2.08−0.08+0.10 M⊙M_{\mathrm{TOV}}^{\mathrm{NPT}} = 2.08^{+0.10}_{-0.08}\,M_{\odot} to MTOVFOPT=2.14−0.10+0.12 M⊙M_{\mathrm{TOV}}^{\mathrm{FOPT}} = 2.14^{+0.12}_{-0.10}\,M_{\odot}.

Because the inferred FOPT onset lies predominantly above ncn_{\mathrm{c}}, the transition leaves no direct imprint on cold and stable neutron stars. The FOPT and NPT posteriors for R1.4R_{1.4} and Λ1.4\Lambda_{1.4} agree at the 68% level for either choice of nLn_{\mathrm{L}}, and their MTOVM_{\mathrm{TOV}} values differ by only ΔMTOV≃0.06\Delta M_{\mathrm{TOV}} \simeq0.06–0.08 M⊙0.08\,M_{\odot}. Another classical signature of an FOPT—a disconnected “twin-star” branch supporting two stable stellar configurations of equal mass but different radii—is likewise strongly disfavored by our analyses. We find that the posterior probability of a twin-star solution under the FOPT hypothesis is ≤0.1%\leq0.1\% for both nL=25n0n_{\mathrm{L}} = 25n_{0} and 12n012n_{0}, in agreement with the model-agnostic exclusion of twin stars reported in Blomqvist et al. [18].

Cold and stable neutron stars thus offer no direct handle on the high-density softening induced by an FOPT, because the transition sets in beyond the densities these stars ever reach. An FOPT remains observable, however, once attention turns to transient systems that momentarily access densities beyond the central density of the most massive stable neutron star. Binary neutron-star post-merger gravitational-wave emission, targeted by next-generation observatories ([65, 4, 30]), offers a natural candidate: the transient remnant briefly attains densities exceeding ncn_{\mathrm{c}}, and an FOPT triggered above ncn_{\mathrm{c}} would imprint itself on the gravitational-wave signal through a shift in the dominant post-merger frequency

and a reduced remnant lifetime ([14, 71, 103, 32]).

Figure 3(a) shows the posterior cs2(n)c_{\mathrm{s}}^{2}(n) obtained with the default dataset d~def\tilde{d}_{\mathrm{def}} for nL=25n0n_{\mathrm{L}} = 25n_{0} under the NPT and FOPT hypotheses. At low densities, cs2c_{\mathrm{s}}^{2} rises under both hypotheses with essentially overlapping posteriors, consistent with the indistinguishable R1.4R_{1.4} and Λ1.4\Lambda_{1.4}. At higher densities, cs2c_{\mathrm{s}}^{2} develops a peak near ∼4n0\sim4n_{0} and decreases before reaching nc≃6n0n_{\mathrm{c}} \simeq6n_{0} in either case. Under the FOPT, both the peak cs2c_{\mathrm{s}}^{2} and its value at ncn_{\mathrm{c}} exceed those under the NPT, reflecting a stiffer neutron-star interior that the pQCD constraint does not overly soften. Beyond the peak, the two posteriors approach the pQCD-predicted value (≃0.32 c2\simeq0.32\,c^{2}) along distinct routes: the NPT descends to a minimum of ∼0.2c2\sim0.2c^{2} near ncn_{\mathrm{c}} and then rises smoothly toward the pQCD-predicted value from below, whereas the FOPT drops to zero across the plateau, rebounds to a smaller post-plateau peak, and approaches the pQCD-predicted value from above. The post-plateau cs2c_{\mathrm{s}}^{2} peak under the FOPT indicates that the matter beyond the FOPT remains markedly non-conformal over an extended window before joining the perturbative regime.

A complementary diagnostic of conformality is the trace anomaly Δ=1/3−p/ε\Delta= 1/3 - p/\varepsilon ([33]), shown in Figure 3(c) for nL=25n0n_{\mathrm{L}} = 25n_{0}. Under the FOPT hypothesis, Δ\Delta first decreases to a dip, then rises monotonically to a pronounced maximum before approaching the perturbative regime from above, whereas the NPT posterior follows the same pattern with a markedly weaker rebound. Moreover, Δ\Delta does not flatten toward zero below nc≃6n0n_{\mathrm{c}} \simeq6 n_{0} under either hypothesis, indicating that the matter inside neutron stars remains non-conformal and strongly coupled.

Image not extracted

Figure 3. Squared sound speed cs2c_{\mathrm{s}}^{2} (top) and trace anomaly Δ=1/3−p/ε\Delta= 1/3 - p/\varepsilon (bottom) as functions of baryon density, with nL=25n0n_{\mathrm{L}} = 25n_{0} in the left column and nL=12n0n_{\mathrm{L}} = 12n_{0} in the right column. Filled bands give the 68% credible intervals and solid lines mark the median values, with FOPT shown in blue and NPT in red. Black dashed lines indicate the conformal references cs2=c2/3c_{\mathrm{s}}^{2} = c^{2}/3 and Δ=0\Delta= 0.

Figure 3(b) and (d) display cs2(n)c_{\mathrm{s}}^{2}(n) and Δ(n)\Delta(n) for nL=12n0n_{\mathrm{L}} = 12 n_{0}. Below ∼5n0\sim5 n_{0}, both panels closely follow their nL=25n0n_{\mathrm{L}} = 25 n_{0} counterparts: cs2c_{\mathrm{s}}^{2} rises to a peak near ∼4n0\sim4 n_{0} that is more pronounced under the FOPT, while Δ\Delta descends toward a dip. At higher densities, the lower extrapolation boundary leaves a clear imprint on the FOPT plateau. The plateau is markedly narrower than for nL=25n0n_{\mathrm{L}} = 25 n_{0}: its median width shrinks from nE−nS=8.51−4.42+3.80n0n_{\mathrm{E}} - n_{\mathrm{S}} = 8.51^{+3.80}_{-4.42} n_{0} at nL=25n0n_{\mathrm{L}} = 25 n_{0} to 3.97−2.28+1.83n03.97^{+1.83}_{-2.28} n_{0} at nL=12n0n_{\mathrm{L}} = 12 n_{0}, so a lower termination density compresses the density window available to the transition. For both choices of nLn_{\mathrm{L}}, Table 2 summarizes the posterior medians and credible intervals of the stellar observables and characteristic densities, the thermodynamic quantities at the transition, together with the Bayes factors.

Table 2. Posterior medians and 68% credible intervals for the maximum nonrotating mass MTOVM_{\mathrm{TOV}} (M⊙M_{\odot}), radius R1.4R_{1.4} (km), tidal deformability Λ1.4\Lambda_{1.4}, and central density ncn_{\mathrm{c}} of the maximum-mass configuration under the NPT and FOPT hypotheses at nL=25n0n_{\mathrm{L}} = 25 n_{0} and nL=12n0n_{\mathrm{L}} = 12 n_{0}. The FOPT onset density nSn_{\mathrm{S}}, FOPT end density nEn_{\mathrm{E}}, baryon-density jump nE−nSn_{\mathrm{E}} - n_{\mathrm{S}}, transition pressure ptp_{\mathrm{t}} (MeV fm−3^{-3}), transition chemical potential μt\mu_{\mathrm{t}} (MeV), transition energy density εt\varepsilon_{\mathrm{t}} (MeV fm−3^{-3}), latent heat Δε\Delta\varepsilon (MeV fm−3^{-3}), and dimensionless latent-heat jump Δε/εt\Delta\varepsilon/\varepsilon_{\mathrm{t}} are reported only in the FOPT columns. The two Bayes factors, BNPTFOPTB_{\mathrm{NPT}}^{\mathrm{FOPT}} (FOPT versus NPT) and BinoutB_{\mathrm{in}}^{\mathrm{out}} (FOPT-out versus FOPT-in), are also listed in the FOPT columns, with their quoted uncertainties given by Monte Carlo standard errors.

QuantitynL=25n0n_{\mathrm{L}} = 25 n_{0} NPTnL=25n0n_{\mathrm{L}} = 25 n_{0} FOPTnL=12n0n_{\mathrm{L}} = 12 n_{0} NPTnL=12n0n_{\mathrm{L}} = 12 n_{0} FOPT
MTOVM_{\mathrm{TOV}}2.07−0.08+0.102.07^{+0.10}_{-0.08}2.15−0.11+0.132.15^{+0.13}_{-0.11}2.08−0.08+0.102.08^{+0.10}_{-0.08}2.14−0.10+0.122.14^{+0.12}_{-0.10}
R1.4R_{1.4}11.79−0.41+0.3711.79^{+0.37}_{-0.41}11.76−0.38+0.3811.76^{+0.38}_{-0.38}11.75−0.39+0.3911.75^{+0.39}_{-0.39}11.79−0.39+0.3911.79^{+0.39}_{-0.39}
Λ1.4\Lambda_{1.4}317−75+86317^{+86}_{-75}309−70+88309^{+88}_{-70}308−70+89308^{+89}_{-70}314−71+92314^{+92}_{-71}
nc/n0n_{\mathrm{c}}/n_{0}6.24−0.57+0.626.24^{+0.62}_{-0.57}6.13−0.85+0.706.13^{+0.70}_{-0.85}6.32−0.57+0.616.32^{+0.61}_{-0.57}5.48−1.08+1.035.48^{+1.03}_{-1.08}
nS/n0n_{\mathrm{S}}/n_{0}—7.35−2.05+4.507.35^{+4.50}_{-2.05}—5.30−1.05+1.765.30^{+1.76}_{-1.05}
nE/n0n_{\mathrm{E}}/n_{0}—17.15−5.19+4.3117.15^{+4.31}_{-5.19}—9.85−2.32+1.519.85^{+1.51}_{-2.32}
(nE−nS)/n0(n_{\mathrm{E}} - n_{\mathrm{S}})/n_{0}—8.51−4.42+3.808.51^{+3.80}_{-4.42}—3.97−2.28+1.833.97^{+1.83}_{-2.28}
ptp_{\mathrm{t}}—626−220+395626^{+395}_{-220}—372−105+163372^{+163}_{-105}
μt\mu_{\mathrm{t}}—1857−190+2731857^{+273}_{-190}—1619−118+1661619^{+166}_{-118}
εt\varepsilon_{\mathrm{t}}—1575−559+13161575^{+1316}_{-559}—999−239+469999^{+469}_{-239}
Δε\Delta\varepsilon—2508−1345+13232508^{+1323}_{-1345}—1043−593+4751043^{+475}_{-593}
Δε/εt\Delta\varepsilon/\varepsilon_{\mathrm{t}}—1.56−1.00+1.201.56^{+1.20}_{-1.00}—1.00−0.64+0.691.00^{+0.69}_{-0.64}
BNPTFOPTB_{\mathrm{NPT}}^{\mathrm{FOPT}}—3.20±0.043.20 \pm0.04—2.84±0.032.84 \pm0.03
BinoutB_{\mathrm{in}}^{\mathrm{out}}—4.01±0.214.01 \pm0.21—1.88±0.071.88 \pm0.07

Table 2.

Finally, we would like to mention that the cs2=0c_{\mathrm{s}}^{2} = 0 plateau under the FOPT follows from the Maxwell construction, in which each pure phase is separately charge-neutral and β\beta-equilibrated. The NPT hypothesis instead yields a smooth cs2(n)c_{\mathrm{s}}^{2}(n) that can in principle describe a continuous crossover from hadronic to quark matter. In addition, an alternative treatment of the phase coexistence is the Gibbs construction, in which charge neutrality is imposed globally and the two phases coexist throughout a mixed phase [34]. The resulting cs2(n)c_{\mathrm{s}}^{2}(n) stays nonzero across the mixed phase and develops kinks at the lower and upper transition densities. Extending our framework to the Gibbs construction is a natural next step toward a more complete characterization of the dense-matter phase transition.

Summary

We have performed Bayesian inference with a non-parametric Gaussian-process EOS for β\beta-equilibrated neutron-star matter, using the GW170817 tidal deformability, the NICER mass–radius measurements of PSR J0740+6620, PSR J0030+0451, PSR J0437−4715, and PSR J0614−3329, ChEFT below 1.5n01.5 n_{0}, and the pQCD sound speed over 25n0≤n≤40n025 n_{0} \leq n \leq40 n_{0} together with the pQCD thermodynamic quantities at npQCD≃40n0n_{\mathrm{pQCD}} \simeq40 n_{0}. Within this framework, we have tested the strong first-order phase transition (FOPT) hypothesis directly against the no strong first-order phase transition (NPT) hypothesis and further tested whether the FOPT onset density lies above (FOPT-out) or below (FOPT-in) the central density of the most massive neutron star.

With the Gaussian-process EOS terminated at nL=25n0n_{\mathrm{L}} = 25 n_{0}, the data moderately favor the FOPT hypothesis over NPT and FOPT-out over FOPT-in, with BNPTFOPT=3.20±0.04B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 3.20 \pm0.04 and Binout=4.01±0.21B_{\mathrm{in}}^{\mathrm{out}} = 4.01 \pm0.21. Lowering the termination density to nL=12n0n_{\mathrm{L}} = 12 n_{0} gives a consistent but weaker preference, with BNPTFOPT=2.84±0.03B_{\mathrm{NPT}}^{\mathrm{FOPT}} = 2.84 \pm0.03 and Binout=1.88±0.07B_{\mathrm{in}}^{\mathrm{out}} = 1.88 \pm0.07. These Bayes factors collectively favor the FOPT hypothesis, with the onset most likely beyond the densities reached inside stable neutron stars. Because the inferred FOPT onset predominantly lies above the central density ncn_{\mathrm{c}}, the canonical-mass observables R1.4R_{1.4} and Λ1.4\Lambda_{1.4} are statistically indistinguishable between the two hypotheses.

Generally, the pQCD constraint disfavors stiff EOSs, in tension with the stiffness required to support a ∼2M⊙\sim2 M_{\odot} neutron star. An FOPT predominantly lying above ncn_{\mathrm{c}} prevents the neutron-star interior from softening excessively under the pQCD constraint and shifts MTOVM_{\mathrm{TOV}} modestly upward, while allowing the EOS soften toward the pQCD regime at higher densities, and thus appears to provide a natural resolution of this tension. Future gravitational-wave observations of binary neutron-star mergers, in particular the post-merger signal accessible to next-generation detectors [65, 4, 30], may further test our present conclusion.

Acknowledgements

We thank Sophia Han and Zhen Zhang for helpful discussions. This work was supported by the National Natural Science Foundation of China under Grant No. 12235010, the National SKA Program of China No. 2020SKA0120300, and the Science and Technology Commission of Shanghai Municipality (Grant No. 23JC1402700). The computations in this paper were run on the Siyuan-1 cluster supported by the Center for High Performance Computing at Shanghai Jiao Tong University.

References

  1. [1]Abbott, B. P., et al. 2017, Phys. Rev. Lett., 119, 161101, doi: 10.1103/PhysRevLett.119.161101DOI
  2. [4]Ackley, K., et al. 2020, Publ. Astron. Soc. Austral., 37, e047, doi: 10.1017/pasa.2020.39DOI
  3. [5]Alford, M. G., Han, S., & Prakash, M. 2013, Phys. Rev. D, 88, 083013, doi: 10.1103/PhysRevD.88.083013DOI
  4. [6]Alford, M. G., Kapustin, A., & Wilczek, F. 1999, Phys. Rev. D, 59, 054502, doi: 10.1103/PhysRevD.59.054502DOI
  5. [7]Alvarez-Castillo, D. E., & Blaschke, D. B. 2017, Phys. Rev. C, 96, 045809, doi: 10.1103/PhysRevC.96.045809
  6. [8]Annala, E., Gorda, T., Hirvonen, J., et al. 2023, Nature Commun., 14, 8451, doi: 10.1038/s41467-023-44051-yDOI
  7. [9]Annala, E., Gorda, T., Kurkela, A., Nättilä, J., & Vuorinen, A. 2020, Nature Phys., 16, 907, doi: 10.1038/s41567-020-0914-9DOI
  8. [10]Aoki, Y., Borsanyi, S., Durr, S., et al. 2009, JHEP, 06, 088, doi: 10.1088/1126-6708/2009/06/088DOI
  9. [11]Aoki, Y., Endrodi, G., Fodor, Z., Katz, S. D., & Szabo, K. K. 2006, Nature, 443, 675, doi: 10.1038/nature05120DOI
  10. [12]Ayriyan, A., Bastian, N. U., Blaschke, D., et al. 2018, Phys. Rev. C, 97, 045802, doi: 10.1103/PhysRevC.97.045802DOI
  11. [13]Ayriyan, A., Ivanytskyi, O., & Blaschke, D. 2025. https://arxiv.org/abs/2509.02554
  12. [14]Bauswein, A., Bastian, N.-U. F., Blaschke, D. B., et al. 2019, Phys. Rev. Lett., 122, 061102, doi: 10.1103/PhysRevLett.122.061102DOI
  13. [15]Baym, G., Pethick, C., & Sutherland, P. 1971, Astrophys. J., 170, 299, doi: 10.1086/151216DOI
  14. [16]Bazavov, A., et al. 2014, Phys. Rev. D, 90, 094503, doi: 10.1103/PhysRevD.90.094503DOI
  15. [17]Blacker, S., Bastian, N.-U. F., Bauswein, A., et al. 2020, Phys. Rev. D, 102, 123023, doi: 10.1103/PhysRevD.102.123023DOI
  16. [18]Blomqvist, S., Ecker, C., Gorda, T., & Vuorinen, A. 2025. https://arxiv.org/abs/2512.19477
  17. [19]Borsanyi, S., Fodor, Z., Hoelbling, C., et al. 2014, Phys. Lett. B, 730, 99, doi: 10.1016/j.physletb.2014.01.007DOI
  18. [20]Brandes, L. W., Weise, W., & Kaiser, N. 2023, Phys. Rev. D, 108, 094014, doi: 10.1103/PhysRevD.108.094014DOI
  19. [21]Cheng, M., et al. 2006, Phys. Rev. D, 74, 054507, doi: 10.1103/PhysRevD.74.054507DOI
  20. [22]Choudhury, D., et al. 2024a, Astrophys. J. Lett., 971, L20, doi: 10.3847/2041-8213/ad5a6fDOI
  21. [24]Christian, J.-E., & Schaffner-Bielich, J. 2020, Astrophys. J. Lett., 894, L8, doi: 10.3847/2041-8213/ab8af4DOI
  22. [25]Christian, J.-E., Schaffner-Bielich, J., & Rosswog, S. 2024, Phys. Rev. D, 109, 063035, doi: 10.1103/PhysRevD.109.063035DOI
  23. [26]Drischler, C., Fünsthal, R. J., Melendez, J. A., & Phillips, D. R. 2020, Phys. Rev. Lett., 125, 202702, doi: 10.1103/PhysRevLett.125.202702DOI
  24. [27]Drischler, C., Hebeler, K., & Schwenk, A. 2019, Phys. Rev. Lett., 122, 042501, doi: 10.1103/PhysRevLett.122.042501DOI
  25. [28]Ecker, C., Jokela, N., & Järvinen, M. 2026, Phys. Rev. D, 113, L041302, doi: 10.1103/x17s-sc9tDOI
  26. [29]Essick, R., Legred, I., Chatzioannou, K., Han, S., & Landry, P. 2023, Phys. Rev. D, 108, 043013, doi: 10.1103/PhysRevD.108.043013DOI
  27. [30]Evans, M., et al. 2021, A Horizon Study for Cosmic Explorer: Science, Observatories, and Community. https://arxiv.org/abs/2109.09882
  28. [31]Freedman, B. A., & McLerran, L. D. 1977, Phys. Rev. D, 16, 1169, doi: 10.1103/PhysRevD.16.1169DOI
  29. [32]Fujimoto, Y., Fukushima, K., Hotokezaka, K., & Kyutoku, K. 2023, Phys. Rev. Lett., 130, 091404, doi: 10.1103/PhysRevLett.130.091404DOI
  30. [33]Fujimoto, Y., Fukushima, K., McLerran, L. D., & Praszalowicz, M. 2022, Phys. Rev. Lett., 129, 252702, doi: 10.1103/PhysRevLett.129.252702DOI
  31. [34]Glendenning, N. K. 1992, Phys. Rev. D, 46, 1274, doi: 10.1103/PhysRevD.46.1274DOI
  32. [35]Gorda, T., Hebeler, K., Kurkela, A., Schwenk, A., & Vuorinen, A. 2023a, Astrophys. J., 955, 100, doi: 10.3847/1538-4357/aceebf
  33. [36]Gorda, T., Komoltsev, O., & Kurkela, A. 2023b, Astrophys. J., 950, 107, doi: 10.3847/1538-4357/acce3aDOI
  34. [37]Gorda, T., Komoltsev, O., Kurkela, A., & Mazeliauskas, A. 2023c, JHEP, 06, 002, doi: 10.1007/JHEP06(2023)002DOI
  35. [38]Gorda, T., Kurkela, A., Paatelainen, R., Säppi, S., & Vuorinen, A. 2021, Phys. Rev. Lett., 127, 162003, doi: 10.1103/PhysRevLett.127.162003DOI
  36. [39]Gorda, T., Kurkela, A., Romatschke, P., Säppi, S., & Vuorinen, A. 2018, Phys. Rev. Lett., 121, 202701, doi: 10.1103/PhysRevLett.121.202701DOI
  37. [40]Grundler, X., & Li, B.-A. 2025, Phys. Rev. D, 112, 103012, doi: 10.1103/hsd4-j54yDOI
  38. [41]Hammond, P., et al. 2026, Phys. Rev. D, 113, 044057, doi: 10.1103/7t3-3t28
  39. [42]Han, S., & Steiner, A. W. 2019, Phys. Rev. D, 99, 083014, doi: 10.1103/PhysRevD.99.083014DOI
  40. [43]Hands, S. 2007, Prog. Theor. Phys. Suppl., 168, 253, doi: 10.1143/PTPS.168.253DOI
  41. [44]Hebeler, K., Lattimer, J. M., Pethick, C. J., & Schwenk, A. 2013, Astrophys. J., 773, 11, doi: 10.1088/0004-637X/773/1/11DOI
  42. [45]Hernandez Vivanco, F., Smith, R., Thrane, E., & Lasky, P. D. 2020a, toast: The neuTrOn stAr STacking package, https://git.ligo.org/francisco.hernandez/toast
  43. [47]Hinderer, T. 2008, Astrophys. J., 677, 1216, doi: 10.1086/533487DOI
  44. [48]Huang, C., & Sourav, S. 2025, Astrophys. J., 983, 17, doi: 10.3847/1538-4357/adbb67DOI
  45. [49]Jeffreys, H. 1939, The Theory of Probability, Oxford Classic Texts in the Physical Sciences
  46. [51]Keller, J., Hebeler, K., & Schwenk, A. 2023, Phys. Rev. Lett., 130, 072701, doi: 10.1103/PhysRevLett.130.072701DOI
  47. [52]Komoltsev, O. 2023, QCD-likelihood-function, https://github.com/OKomoltsev/QCD-likelihood-functionDOI
  48. [54]Komoltsev, O., Gorda, T., & Kurkela, A. 2025, Marginalized QCD likelihood function, v3, Zenodo, doi: 10.5281/zenodo.15407795
  49. [55]Komoltsev, O., & Kurkela, A. 2022, Phys. Rev. Lett., 128, 202701, doi: 10.1103/PhysRevLett.128.202701DOI
  50. [56]Komoltsev, O., Somasundaram, R., Gorda, T., et al. 2024, Phys. Rev. D, 109, 094030, doi: 10.1103/PhysRevD.109.094030DOI
  51. [57]Kong, A., Liu, J. S., & Wong, W. H. 1994, Journal of the American Statistical Association, 89, 278, doi: 10.1080/01621459.1994.10476469
  52. [58]Kumar, A., Thapa, V. B., & Sinha, M. 2023, Phys. Rev. D, 107, 063024, doi: 10.1103/PhysRevD.107.063024DOI
  53. [59]Kurkela, A., Romatschke, P., & Vuorinen, A. 2010, Phys. Rev. D, 81, 105021, doi: 10.1103/PhysRevD.81.105021DOI
  54. [60]Kurkela, A., & Vuorinen, A. 2016, Phys. Rev. Lett., 117, 042501, doi: 10.1103/PhysRevLett.117.042501DOI
  55. [61]Lee, M. D., & Wagenmakers, E.-J. 2014, Bayesian Cognitive Modeling: A Practical Course (Cambridge University Press)
  56. [63]Lindblom, L., Lewis, S. M., & Weber, F. 2025, Phys. Rev. D, 111, 123035, doi: 10.1103/j1z7-jfc6DOI
  57. [64]Lynn, J. E., Tews, I., Carlson, J., et al. 2016, Phys. Rev. Lett., 116, 062501, doi: 10.1103/PhysRevLett.116.062501DOI
  58. [65]Maggiore, M., et al. 2020, JCAP, 03, 050, doi: 10.1088/1475-7516/2020/03/050DOI
  59. [66]Mauviard, L., et al. 2025, Astrophys. J., 995, 60, doi: 10.3847/1538-4357/ae145dDOI
  60. [68]Miller, M. C., et al. 2019, Astrophys. J. Lett., 887, L24, doi: 10.3847/2041-8213/ab50c5DOI
  61. [70]Montana, G., Tolos, L., Hanauske, M., & Rezzolla, L. 2019, Phys. Rev. D, 99, 103009, doi: 10.1103/PhysRevD.99.103009DOI
  62. [71]Most, E. R., Papenfort, L. J., Dexheimer, V., et al. 2019, Phys. Rev. Lett., 122, 061101, doi: 10.1103/PhysRevLett.122.061101DOI
  63. [72]Negele, J. W., & Vautherin, D. 1973, Nucl. Phys. A, 207, 298, doi: 10.1016/0375-9474(73)90349-7DOI
  64. [73]Oppenheimer, J. R., & Volkoff, G. M. 1939, Phys. Rev., 55, 374, doi: 10.1103/PhysRev.55.374DOI
  65. [74]Pang, P. T. H., Dietrich, T., Tews, I., & Van Den Broeck, C. 2020, Phys. Rev. Res., 2, 033514, doi: 10.1103/PhysRevResearch.2.033514DOI
  66. [75]Pang, P. T. H., Tews, I., Coughlin, M. W., et al. 2021, Astrophys. J., 922, 14, doi: 10.3847/1538-4357/ac19abDOI
  67. [76]Postnikov, S., Prakash, M., & Lattimer, J. M. 2010, Phys. Rev. D, 82, 024016, doi: 10.1103/PhysRevD.82.024016DOI
  68. [77]Rasmussen, C. E., & Williams, C. K. I. 2006, Gaussian Processes for Machine Learning (Cambridge, MA: MIT Press)DOI
  69. [79]Riley, T. E., et al. 2019, Astrophys. J. Lett., 887, L21, doi: 10.3847/2041-8213/ab481cDOI
  70. [81]Saha, A. K., Malik, T., & Mallick, R. 2025, doi: 10.3847/1538-4357/ae7a44DOI
  71. [82]Saha, A. k., & Mallick, R. 2024. https://arxiv.org/abs/2410.21877
  72. [83]Salmi, T., et al. 2024a, Astrophys. J., 974, 294, doi: 10.3847/1538-4357/ad5f1fDOI
  73. [85]Scott, D. W. 1992, Multivariate Density Estimation: Theory, Practice, and Visualization (John Wiley & Sons), doi: 10.1002/9780470316849DOI
  74. [86]Seidov, Z. F. 1971, Soviet Astronomy, 15, 347
  75. [87]Sieniawska, M., Turczanski, W., Bejger, M., & Zdunik, J. L. 2019, Astron. Astrophys., 622, A174, doi: 10.1051/0004-6361/201833969DOI
  76. [88]Somasundaram, R., Tews, I., & Margueron, J. 2023, Phys. Rev. C, 107, 025801, doi: 10.1103/PhysRevC.107.025801DOI
  77. [89]Steiner, A. W., Heinke, C. O., Bogdanov, S., et al. 2018, Mon. Not. Roy. Astron. Soc., 476, 421, doi: 10.1093/mnras/sty215DOI
  78. [90]Steiner, A. W., Lattimer, J. M., & Brown, E. F. 2013, Astrophys. J. Lett., 765, L5, doi: 10.1088/2041-8205/765/1/L5DOI
  79. [91]Takatsy, J., Kovacs, P., Wolf, G., & Schaffner-Bielich, J. 2023, Phys. Rev. D, 108, 043002, doi: 10.1103/PhysRevD.108.043002DOI
  80. [92]Tan, H. D., Dore, T., Dexheimer, V., Noronha-Hostler, J., & Yunes, N. 2022, Phys. Rev. D, 105, 023018, doi: 10.1103/PhysRevD.105.023018DOI
  81. [93]Tang, S.-P., Huang, Y.-J., & Fan, Y.-Z. 2025, Phys. Rev. D, 112, 083009, doi: 10.48550/bmsk-8n85
  82. [94]Tang, S.-P., Jiang, J.-L., Gao, W.-H., Fan, Y.-Z., & Wei, D.-M. 2021, Phys. Rev. D, 103, 063026, doi: 10.1103/PhysRevD.103.063026DOI
  83. [95]Tews, I., Krüger, T., Hebeler, K., & Schwenk, A. 2013, Phys. Rev. Lett., 110, 032504, doi: 10.1103/PhysRevLett.110.032504DOI
  84. [96]Thrane, E., & Talbot, C. 2019, Publ. Astron. Soc. Austral., 36, e010, doi: 10.1017/pasa.2019.2DOI
  85. [97]Tolman, R. C. 1939, Phys. Rev., 55, 364, doi: 10.1103/PhysRev.55.364DOI
  86. [98]Trotta, R. 2008, Contemp. Phys., 49, 71, doi: 10.1080/00107510802066753DOI
  87. [99]Tsaloukidis, L., Koliogiannis, P. S., Kanakis-Pegios, A., & Moustakidis, C. C. 2023, Phys. Rev. D, 107, 023012, doi: 10.1103/PhysRevD.107.023012DOI
  88. [100]Verma, A., Saha, A. K., & Mallick, R. 2025, Astrophys. J., 985, 1, doi: 10.3847/1538-4357/adcee0DOI
  89. [101]Vinciguerra, S., et al. 2023, An updated mass-radius analysis of the 2017-2018 NICER data set of PSR J0030+0451, Zenodo, doi: 10.5281/zenodo.8239000DOI
  90. [103]Weih, L. R., Hanauske, M., & Rezzolla, L. 2020, Phys. Rev. Lett., 124, 171103, doi: 10.1103/PhysRevLett.124.171103DOI
  91. [104]Zhou, W., Shen, H., Hu, J., & Zhang, Y. 2024, Phys. Rev. D, 110, 043017, doi: 10.1103/PhysRevD.110.043017DOI

Paper details

vlnc:W_eubu7t3k
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents