betaFeedback

WorksPhysics and Astronomy

On the Stability of Tolman Oppenheimer Volkoff Fireholes

Abstract

We investigate the radial stability of singular Tolman–Oppenheimer–Volkoff fireholes with a radiation equation of state. These configurations contain a naked, negative point-mass singularity surrounded by a positive-density fluid, whose mass profile develops a Schwarzschild-like plateau before approaching an extended, self-similar atmosphere. Using the Prüfer phase method to identify the fundamental radial mode, we consider two outer boundary conditions: vanishing Lagrangian pressure perturbation (open firehole) and vanishing radial displacement (boxed firehole). For both, the fundamental squared frequency changes from positive to negative as the outer boundary exceeds a critical radius RcR_c. We find that RcR_c follows approximately the same scaling with the central mass parameter as the equilibrium firehole radius RR. However, the stability thresholds differ significantly: RcO<RR_c^{\rm O}<R for the open firehole, whereas RcB≃RR_c^{\rm B}\simeq R for the boxed one. The regular zero mode is the derivative of the equilibrium family with respect to the inner lapse normalization at fixed central mass. This gives the boxed threshold a turning-point interpretation and, together with a scaling symmetry of the TOV equations, implies RcB→RR_c^{\rm B}\to R in the limit of large central mass. Variational bounds show that the rigid wall is the most stabilizing separated, self-adjoint outer boundary condition and that the unbounded configuration is unstable. Tolman–Oppenheimer–Volkoff fireholes are therefore radially unstable unless confined within approximately their firehole radius.

Introduction

The Tolman–Oppenheimer–Volkoff (TOV) equations provide the fundamental description of static, spherically symmetric self-gravitating fluids in general relativity [1, 2]. For a specified equation of state, their solutions determine the equilibrium structure of relativistic stars and the relation between their gravitational mass, radius, and internal matter distribution. They have consequently played a central role in relativistic stellar astrophysics, most prominently in models of neutron stars, but also in relativistic treatments of white dwarfs and other compact stellar configurations. Much of our understanding of the maximum masses, compactness, and stability of relativistic stars ultimately rests on properties of solutions of the TOV equations.

The equilibrium solutions also provide the starting point for studying the response and dynamics of relativistic stars. Radial perturbations lead to the relativistic stellar pulsation equations developed by Chandrasekhar [3, 4], whose eigenvalues determine the onset of radial instability. Nonradial perturbations characterize stellar oscillations and tidal deformability, including the tidal Love numbers and the approximate universal I–Love–Q relations [5, 6]. At the fully nonlinear level, TOV solutions and their generalizations also form the building blocks for initial data and numerical-relativity simulations of isolated and binary compact stars [7, 8, 9].

Nearly all these applications, however, concern the regular branch of TOV solutions, for which the density and pressure remain finite at the origin and the enclosed mass vanishes as r→0r\to0. Much less attention has been given to singular solutions. A particularly intriguing example was identified by Zurek and Page [10], and later considered by ’t Hooft [11], in studies of self-gravitating thermal radiation. Integrating the TOV equations inward from a radiation atmosphere, Zurek and Page found solutions with positive fluid density everywhere but no event horizon. Instead, these configurations develop a high-density region near the nominal Schwarzschild radius and approach, at small radii, a geometry dominated by a negative point mass. Their mass profiles exhibit an approximately Schwarzschild-like plateau, suggesting a compact central object surrounded by an extended radiation atmosphere. Zurek and Page emphasized the thermodynamic properties of these configurations, including the possibility that the strongly blueshifted radiation carries an entropy comparable to the Bekenstein–Hawking entropy.

Despite these remarkable properties, the dynamical stability of the singular branch, which we refer to as TOV fireholes (TOVFs), remains largely unexplored. Anastopoulos and Savvidou [12] classified singular TOV solutions, showed that near the center they are locally isomorphic to negative-mass Schwarzschild, argued that the strongly repulsive singularity stabilizes the surrounding matter, and undertook a preliminary investigation of their dynamical stability. Here we show that, despite the repulsive core, TOVFs are radially unstable unless the fluid is confined within approximately the firehole radius. The existence of a static equilibrium solution does not establish its stability under perturbations. Moreover, unlike ordinary relativistic stars, TOVFs with a radiation equation of state have no finite radius at which the pressure vanishes. Their fluid atmosphere extends indefinitely and approaches a self-similar, non-asymptotically flat configuration. Consequently, the radial pulsation problem requires particular care in specifying the inner regularity condition and the outer boundary condition. An important question is whether the compact firehole can support stable radial oscillations independently of its extended atmosphere, or whether stability requires a mechanism that confines the surrounding fluid.

In this work, we investigate the radial stability of TOVFs with a radiation equation of state, p=ρ/3p = \rho/3, using Chandrasekhar’s relativistic pulsation equations. We consider two boundary-value problems at a prescribed outer radius RoutR_{\mathrm{out}}: vanishing Lagrangian pressure perturbation, which we call TOVF Open (TOVFO), and vanishing radial displacement at a rigid boundary, which we call TOVF Boxed (TOVFB). Using the Prüfer phase method, we identify the fundamental radial mode and determine the critical radius RcR_{\mathrm{c}} at which its squared frequency changes sign. We find that the critical radii approximately follow the same scaling with the central negative mass parameter as the characteristic firehole radius RR and plateau mass MM.

Our principal result is that the stability threshold depends significantly on the outer boundary condition. For TOVFO, the fundamental mode becomes unstable at a critical radius RcO<RR_{\mathrm{c}}^{\mathrm{O}} < R, whereas for TOVFB, the threshold nearly coincides with the firehole radius, RcB≃RR_{\mathrm{c}}^{\mathrm{B}} \simeq R. A variational analysis establishes the stabilizing effect of the rigid boundary and explains the ordering of these critical radii. We show that the regular zero-frequency mode is the displacement connecting neighboring equilibria, which explains why RcBR_{\mathrm{c}}^{\mathrm{B}} approaches RR. Since even the boxed configuration, which is the most stabilizing separated, self-adjoint outer boundary condition, develops a radial instability when the fluid domain extends beyond approximately the firehole radius, TOVFs cannot be regarded as radially stable isolated objects without an additional mechanism that confines or otherwise stabilizes their extended atmosphere. A variational argument extends this conclusion to the complete, unbounded atmosphere.

The paper is organized as follows. In Sec. II, we construct the TOVF equilibrium solutions, describe their interior and exterior structure, and examine the scaling relations governing their characteristic mass and radius. In Sec. III, we formulate the radial pulsation problem, determine the stability thresholds for open and boxed boundary conditions, investigate their eigenfunctions and variational properties, relate the zero modes to neighboring equilibria, and derive bounds valid for general confinement and for the unbounded atmosphere. We summarize our results and discuss their implications in Sec. IV.

TOV Fireholes

A static, spherically symmetric spacetime metric may be written as

ds2=−e2ν(r)dt2+e2λ(r)dr2+r2dΩ2,(1)ds^2 = -e^{2\nu(r)}dt^2 + e^{2\lambda(r)}dr^2 + r^2 d\Omega^2, \tag*{(1)}

where dΩ2=dθ2+sin⁡2θ dϕ2d\Omega^2 = d\theta^2 + \sin^2\theta\,d\phi^2. If the matter is a perfect fluid, the Einstein equations, in G=c=1G = c = 1 units, yield the so-called TOV equations [2, 13]

m′=4πr2ρ,(2)m' = 4\pi r^2 \rho, \tag*{(2)}
p′=−(ρ+p)(m+4πr3p)r(r−2m),(3)p' = -\frac{(\rho+ p)(m + 4\pi r^3 p)}{r(r - 2m)}, \tag*{(3)}
ν′=−p′ρ+p,(4)\nu' = -\frac{p'}{\rho+ p}, \tag*{(4)}

where primes denote derivatives with respect to the radial coordinate and e−2λ≡1−2m/re^{-2\lambda} \equiv1 - 2m/r. In these equations, ρ\rho is the total mass-energy density, pp is the pressure, and mm is the effective mass enclosed within the radius rr.

As in the Zurek and Page [10] study, we will focus on the Γ\Gamma-law equation of state (EOS) p=(Γ−1)ρp = (\Gamma- 1)\rho with Γ=4/3\Gamma= 4/3. Figure 1 reproduces Fig. 2 of Ref. [10]. The integration of the Tolman–Oppenheimer–Volkoff (TOV) equations started at a radius 1,265 with m=12.65m = 12.65 and ρ=1.5×10−10\rho= 1.5 \times10^{-10} and proceeded inwards. From the behavior of the mass function mm in the top panel, we observe four domains: an interior core where the mass function is approximately constant, with m≃−μ<0m \simeq-\mu< 0; a transition layer where mm grows rapidly when the density reaches a peak; and a Schwarzschild-like plateau where m≃Mm \simeq M that ends with a linear self-similar asymptotic tail. It is important to emphasize that with a Γ\Gamma-law EOS, the density, and thus the pressure, never vanishes at a finite radius. There is no surface radius. This issue will play an essential role in the radial stability analysis. Furthermore, with the metric of the form in Eq. (1), the presence of a horizon is signaled by the location where 1−2m/r1 - 2m/r vanishes. This quantity is displayed in the bottom panel of Figure 1. For the singular solutions considered here and the one in Figure 1, the quantity 1−2m(r)/r1 - 2m(r)/r remains positive: the negative-mass singularity is not enclosed by a horizon.

Zurek–Page singular solution to the TOV equations

Figure 1. Zurek–Page singular solution to the TOV equations, as in Fig. 2 of Ref. [10]. The outer radius was at 1,265 with m=12.65m = 12.65 and ρ=1.5×10−10\rho= 1.5 \times10^{-10}. From top to bottom are the mass function mm, the density ρ\rho, and the quantity 1−2m/r1 - 2m/r. The dashed line in the top panel was obtained from −m(r)×10−4-m(r) \times10^{-4}.

With Γ=4/3\Gamma= 4/3, Eq. (4) yields −4ν′=(ln⁡ρ)′-4\nu' = (\ln\rho)'. Thus, 4πρe4ν=K4\pi\rho e^{4\nu}=K, with KK an integration constant. We introduce the definition y≡e2νy\equiv e^{2\nu}, so the TOV equations become

m′=K(ry)2(5)m'=K\left(\frac{r}{y}\right)^2 \tag*{(5)}
y′=(yr)[2mr+2K3(ry)2](1−2mr)−1.(6)y'=\left(\frac{y}{r}\right)\left[\frac{2m}{r}+\frac{2K}{3}\left(\frac{r}{y}\right)^2\right]\left(1-\frac{2m}{r}\right)^{-1}. \tag*{(6)}

Without loss of generality, we set K=1K=1, which fixes the overall scaling associated with the first integral of Eq. (4). The singular solutions nevertheless retain a second integration constant, denoted below by AA, which fixes the normalization of the inner lapse and must be specified when defining a one-parameter family. Other solutions can be obtained with the appropriate rescaling.

Close to the origin, in the interior core, Eq. (6) reads y′≃−y/ry'\simeq-y/r, and with Eq. (5), we get that

y≃Ar(7)y\simeq\frac{A}{r} \tag*{(7)}
m≃−μ+r55A2,(8)m\simeq-\mu+\frac{r^5}{5A^2}, \tag*{(8)}

where A>0A>0 is the remaining integration constant. The second expression can be rewritten as

mμ≃−1+(rR∗)5,(9)\frac{m}{\mu}\simeq-1+\left(\frac{r}{R_*}\right)^5, \tag*{(9)}

where R∗5≡5μA2R_*^5\equiv5\mu A^2. This scale is defined by extrapolating the inner expansion. It need not coincide with the numerically determined radius where m(r)=0m(r)=0. R∗R_* is the natural scale at which the positive fluid mass becomes comparable to the negative point mass. It also shows that the inner expansion depends on rr through r/R∗r/R_*.

The transition layer is characterized by the rapid growth of the mass function and by a minimum of e−2λ=1−2m/re^{-2\lambda}=1-2m/r, after which the mass function approaches a Schwarzschild-like plateau. We define the characteristic firehole radius RR as the location of this minimum, or equivalently, the maximum of e2λe^{2\lambda}. The corresponding mass M=m(R)M=m(R) defines the firehole mass. Unlike the surface of an ordinary star, RR is not associated with vanishing pressure, but is determined entirely by the equilibrium geometry. By construction, (m/r)′∣R=0\left.(m/r)'\right|_R=0, which implies m′(R)=M/Rm'(R)=M/R, and 4πR3ρ(R)=M4\pi R^3\rho(R)=M. This geometrical definition of RR provides a natural reference scale for the critical radii obtained from the radial stability analysis.

From Figure 1, at large radius, the numerical solutions approach a linear self-similar asymptotic tail, suggesting m=arm=ar. This requires from Eq. (5) that (r/y)2=a(r/y)^2=a, and hence Eq. (6) yields

1=2a+2a/31−2a.(10)1=\frac{2a+2a/3}{1-2a}. \tag*{(10)}

Thus, a=3/14a=3/14. Therefore, the asymptotic solutions are

y≃143 r(11)y\simeq\sqrt{\frac{14}{3}}\,r \tag*{(11)}
m≃314 r.(12)m\simeq\frac{3}{14}\,r. \tag*{(12)}

Thus, the solutions approach a common self-similar asymptotic branch. Furthermore, since y=e2νy=e^{2\nu} grows linearly with rr, these spacetimes are not asymptotically flat.

We parametrize the second integration constant AA, associated with the normalization of the inner lapse, as

A=μq5,(13)A=\frac{\mu^q}{\sqrt{5}}, \tag*{(13)}

where the numerical factor is introduced for convenience. Figure 2 shows solutions for the case q=0q=0. The top panel shows ∣m∣|m| and the bottom panel y=e2νy=e^{2\nu} as functions of r/Rr/R. Figure 3 shows the corresponding solutions for the case q=1/3q=1/3. In both families, the mass function exhibits the same qualitative structure: a negative mass interior core, a rapid transition layer, an approximately Schwarzschild-like plateau, and an extended self-similar atmosphere. The principal difference between the families is the scaling of the characteristic length and mass scales with μ\mu.

Solutions in the range $10^2\leq\mu\leq10^9$

Figure 2. Solutions in the range 102≤μ≤10910^2\leq\mu\leq10^9, using as initial conditions Eqs. (7) and (8) with q=0q=0. The top panel shows ∣m∣|m| and the bottom panel y=e2νy=e^{2\nu}, as functions of r/Rr/R.

Solutions for $q=1/3$

Figure 3. Similar to Figure 2 but with q=1/3q = 1/3.

An enlightening form of displaying results is by following ’t Hooft [11] and introducing the variables

Z≡e2λ=(1−2mr)−1(14)Z\equiv e^{2\lambda}=\left(1-\frac{2m}{r}\right)^{-1} \tag*{(14)}
X≡e2νr=yr.(15)X\equiv\frac{e^{2\nu}}{r}=\frac{y}{r}. \tag*{(15)}

Figure 4 and Figure 5 display in the variables ZZ and XX the solutions of Figure 2 and Figure 3, respectively. The transition from the negative-mass interior to the Schwarzschild-like plateau is particularly evident in these variables. At large radii, all solutions approach the same self-similar asymptotic values, Z→7/4Z \to7/4 and X→14/3X \to\sqrt{14/3}.

Singular solutions in terms of variables $Z$ and $X$

Figure 4. Singular solutions from Figure 2 in terms of variables ZZ and XX.

Singular solutions in terms of variables $Z$ and $X$

Figure 5. Singular solutions from Figure 3 in terms of variables ZZ and XX.

We now examine how the characteristic firehole radius RR and mass MM depend on the central negative mass parameter μ\mu. With AA given by Eq. (13), the characteristic scale

R∗5=5μA2R_*^5 = 5\mu A^2

becomes

R∗=μ(1+2q)/5.(16)R_* = \mu^{(1+2q)/5}. \tag*{(16)}

To quantify this behavior, we introduce the dimensionless variables

r^=rR∗,m^=mR∗.(17)\hat{r} = \frac{r}{R_*}, \qquad\hat{m} = \frac{m}{R_*}. \tag*{(17)}

The firehole radius and mass can then be written as

R=R^R∗,M=M^R∗,(18)R = \hat{R}R_*, \qquad M = \hat{M}R_*, \tag*{(18)}

where R^=R/R∗\hat{R} = R/R_* and M^=M/R∗\hat{M} = M/R_*. Numerically, these dimensionless quantities vary only weakly with μ\mu and approach constant values over the large-μ\mu range examined. Denoting these limiting values by R^∞\hat{R}_{\infty} and M^∞\hat{M}_{\infty}, we obtain

R≃R^∞μ(1+2q)/5,M≃M^∞μ(1+2q)/5.(19)\begin{aligned} R &\simeq\hat{R}_{\infty}\mu^{(1+2q)/5},\\ M &\simeq\hat{M}_{\infty}\mu^{(1+2q)/5}. \tag*{(19)} \end{aligned}

Thus, the predicted power-law exponent for these families is

p∗(q)=1+2q5.(20)p_*(q) = \frac{1+2q}{5}. \tag*{(20)}

The scaling of R∗R_* follows directly from the inner asymptotic expansion, whereas the approximately constant ratios R/R∗R/R_* and M/R∗M/R_* are established numerically. As shown in Sec. III C, the limiting values do not depend on qq; numerically, R^∞≃1.2457\hat{R}_{\infty} \simeq1.2457 and M^∞≃0.6229\hat{M}_{\infty} \simeq0.6229.

The numerical results support these relations. The power-law fits listed in Table I give, for q=0q = 0, R≃1.311μ0.196R \simeq1.311\mu^{0.196} and M≃0.614μ0.201M \simeq0.614\mu^{0.201}, with fitted exponents close to the predicted value p∗(0)=1/5p_*(0) = 1/5; the same agreement holds for the other values of qq in the table. The compactness of the firehole approaches the Schwarzschild value: M/R≃0.482M/R \simeq0.482 at μ=102\mu= 10^2 for q=0q = 0, and M/R→1/2M/R \to1/2 as μ→∞\mu\to\infty for q<2q < 2, so that the minimum of 1−2m/r1 - 2m/r at RR tends to zero, although the solutions contain no event horizon at any finite μ\mu.

The choice q=1/3q = 1/3 has a particularly simple interpretation. Near the origin, the lapse satisfies e2ν≃A/re^{2\nu} \simeq A/r. Expressing this relation in terms of r^=r/R∗\hat{r} = r/R_* gives

e2ν≃15r^μ(3q−1)/5.(21)e^{2\nu} \simeq\frac{1}{\sqrt{5}\hat{r}}\mu^{(3q-1)/5}. \tag*{(21)}

Hence, q=1/3q = 1/3 is the unique choice for which the leading inner lapse profile, expressed in terms of r^\hat{r}, becomes independent of μ\mu. This provides an interpretation of the μ1/3\mu^{1/3} scaling associated with the Zurek–Page family: it corresponds to comparing solutions with a common dimensionless inner lapse normalization.

More generally, the TOV equations do not determine a unique relationship between the central negative mass μ\mu and the characteristic firehole mass MM. Such a relationship arises only after specifying how the remaining integration constant AA is chosen along a family of solutions.

Up to the scaling symmetry discussed in Sec. III C, the singular solutions form a one-parameter family labeled by β=μ/R∗\beta= \mu/R_\ast, and the choice of qq only fixes how β\beta grows with μ\mu.

Radial Stability Analysis

We study radial oscillations using the formulation of Kokkotas and Ruoff [14], based on Chandrasekhar’s original treatment [3, 4].

Starting with the time-dependent radial displacement δr(t,r)\delta r(t,r) of a fluid element located at position rr, one introduces a harmonic time dependence of the form

δr(t,r)=ξ(r)eiωt=ζ(r)eνeiωt/r2.(22)\delta r(t,r) = \xi(r)e^{i\omega t} = \zeta(r)e^{\nu}e^{i\omega t}/r^2 . \tag*{(22)}

The master equation for ζ\zeta reads [14]

(Pζ′)′+(Q+ω2W)ζ=0,(23)(P\zeta')' + (Q+\omega^2W)\zeta= 0 , \tag*{(23)}

which, for convenience, one rewrites as

ζ′=ηP,(24)\zeta' = \frac{\eta}{P}, \tag*{(24)}
η′=−(Q+ω2W)ζ.(25)\eta' = -(Q+\omega^2W)\zeta. \tag*{(25)}

The coefficients are

r2W=(ρ+p)e3λ+ν,(26)r^2W = (\rho+p)e^{3\lambda+\nu}, \tag*{(26)}
r2P=γpeλ+3ν,(27)r^2P = \gamma p e^{\lambda+3\nu}, \tag*{(27)}
r2Q=(ρ+p)eλ+3ν[ν′2+4ν′r−8πpe2λ],(28)r^2Q = (\rho+p)e^{\lambda+3\nu}\left[\nu'^2+\frac{4\nu'}{r}-8\pi p e^{2\lambda}\right], \tag*{(28)}

where

γ=ρ+ppdpdρ.(29)\gamma= \frac{\rho+p}{p}\frac{dp}{d\rho}. \tag*{(29)}

With our EOS p=ρ/3p=\rho/3 and our normalization 4πρe4ν=14\pi\rho e^{4\nu}=1, these coefficients can be rewritten as

r2W=e3(λ−ν),(30)r^2W = e^{3(\lambda-\nu)}, \tag*{(30)}
r2P=13eλ−ν,(31)r^2P = \frac{1}{3}e^{\lambda-\nu}, \tag*{(31)}
r2Q=eλ−ν[(ν′)2+4ν′r−23e2λ−4ν],(32)r^2Q = e^{\lambda-\nu}\left[(\nu')^2+\frac{4\nu'}{r}-\frac{2}{3}e^{2\lambda-4\nu}\right], \tag*{(32)}

where we have absorbed a factor of 3π3\pi in the coefficients.

We solve Eqs. (24)–(25) on [R0,Rout][R_0,R_{\rm out}] as an eigenvalue problem for ω2\omega^2. The radiation fluid has p(Rout)>0p(R_{\rm out})>0 at every finite RoutR_{\rm out}, so RoutR_{\rm out} is a chosen boundary of the perturbation problem rather than the surface of a star matched directly to a vacuum. Truncating the equilibrium at RoutR_{\rm out} therefore presupposes an exterior agent that supplies the pressure p(Rout)p(R_{\rm out}). The two outer conditions introduced below correspond to two idealized agents: a constant-pressure bath, which leaves the Lagrangian pressure of the boundary fluid element unchanged, and a rigid wall held at fixed areal radius RoutR_{\rm out}, which prevents the boundary element from moving.

The initial conditions at R0R_0 are derived as follows. We found in Sec. II that as r→0r\to0, m→−μm\to-\mu and

e2ν≃Ar(33)e^{2\nu}\simeq\frac{A}{r} \tag*{(33)}
e2λ≃(1+2μr)−1≃r2μ.(34)e^{2\lambda}\simeq\left(1+\frac{2\mu}{r}\right)^{-1}\simeq\frac{r}{2\mu}. \tag*{(34)}

Notice that with our choice A=μq/5A=\mu^q/\sqrt{5}

eλ−ν=Er,(35)e^{\lambda-\nu}=Er, \tag*{(35)}

where

E−2=2μA=45μq+1.(36)E^{-2}=2\mu A=\sqrt{\frac{4}{5}\mu^{q+1}}. \tag*{(36)}

With these approximations, the coefficients WW, PP, and QQ read

W≃rE3(37)W\simeq rE^3 \tag*{(37)}
P≃E3r(38)P\simeq\frac{E}{3r} \tag*{(38)}
Q≃−7E4r3.(39)Q\simeq-\frac{7E}{4r^3}. \tag*{(39)}

Substitution of these approximations into Eq. (23) yields

(ζ′r)′−214r3ζ=0,(40)\left(\frac{\zeta'}{r}\right)'-\frac{21}{4r^3}\zeta=0, \tag*{(40)}

where we have neglected the term with the WW coefficient because W=O(r)W=O(r) compared with Q=O(r−3)Q=O(r^{-3}). Hence,

ζ=c1r7/2+c2r−3/2.(41)\zeta=c_1r^{7/2}+c_2r^{-3/2}. \tag*{(41)}

Recall that δr=eνζ/r2\delta r=e^\nu\zeta/r^2. Since eν∼r−1/2e^\nu\sim r^{-1/2}, the regular branch ζ∼r7/2\zeta\sim r^{7/2} gives δr∼r\delta r\sim r, whereas the second branch ζ∼r−3/2\zeta\sim r^{-3/2} gives δr∼r−4\delta r\sim r^{-4}. The latter, therefore, corresponds to a divergent physical displacement and must be discarded. Hence, the physically relevant solution is ζ=c1r7/2\zeta=c_1r^{7/2}. The same selection follows from the natural Sturm–Liouville norm. Since W∼rW\sim r near the origin, the regular branch satisfies Wζ2∼r8W\zeta^2\sim r^8, whereas the singular branch gives Wζ2∼r−2W\zeta^2\sim r^{-2}. Only the former is integrable at r=0r=0. The singular endpoint is therefore of limit-point type, so no additional independent boundary condition is required there beyond selecting the admissible solution.

Since the radial perturbation equation is defined up to a rescaling constant on ζ\zeta, we will also set c1=1c_1=1. Finally from Eq. (24),

η=ζ′P=7E6r3/2.(42)\eta=\zeta'P=\frac{7E}{6}r^{3/2}. \tag*{(42)}

The initial conditions at R0R_0 will then be

ζ=R07/2(43)\zeta=R_0^{7/2} \tag*{(43)}
η=ζ′P=7E6R03/2.(44)\eta=\zeta'P=\frac{7E}{6}R_0^{3/2}. \tag*{(44)}

For the outer boundary, we will consider two situations: a TOVF with an open boundary, referred to as TOVFO, and a TOVF with a rigid wall, referred to as TOVFB. For the TOVFO, we impose η(Rout)=0\eta(R_{\mathrm{out}})=0, equivalently Δp(Rout)=0\Delta p(R_{\mathrm{out}})=0, and for the TOVFB, we require that ζ(Rout)=0\zeta(R_{\mathrm{out}})=0.

At each outer radius RoutR_{\mathrm{out}}, we follow the nodeless, fundamental radial mode. We denote by RcR_{\mathrm{c}} the critical radius at which this mode has ω2=0\omega^2=0; its eigenvalue is positive below and negative above this crossing in the range examined. For the boxed problem this crossing is unique for every RoutR_{\mathrm{out}}, because ω0,B2\omega_{0,\mathrm{B}}^2 decreases monotonically with RoutR_{\mathrm{out}} (Sec. III D).

To determine the fundamental eigenvalue, we employ the Prüfer phase method, which provides a convenient way to identify the radial eigenmodes by their number of nodes. This is particularly important because a direct shooting method that enforces the outer boundary condition does not, by itself, distinguish the fundamental mode from higher overtones.

We introduce the Prüfer variables b(r)b(r) and θ(r)\theta(r) through

ζ=bsin⁡θ,η=κbcos⁡θ,(45)\zeta=b\sin\theta,\qquad\eta=\kappa b\cos\theta, \tag*{(45)}

where κ>0\kappa>0 is an arbitrary constant that can be chosen to improve numerical conditioning. Substituting these expressions into Eqs. (24) and (25), we obtain

θ′=κPcos⁡2θ+Q+ω2Wκsin⁡2θ.(46)\theta'=\frac{\kappa}{P}\cos^2\theta+\frac{Q+\omega^2W}{\kappa}\sin^2\theta. \tag*{(46)}

The phase equation is independent of the amplitude bb, so the eigenvalues can be determined without integrating the potentially rapidly varying displacement amplitude.

The regular inner solution, Eqs. (43) and (44), determines the initial phase,

θ(R0)=arctan⁡(6κR027E).(47)\theta(R_0)=\arctan\left(\frac{6\kappa R_0^2}{7E}\right). \tag*{(47)}

The two outer boundary conditions become phase conditions. For TOVFO, η(Rout)=0\eta(R_{\mathrm{out}})=0 requires

θ(Rout)=π2+nπ,(48)\theta(R_{\mathrm{out}})=\frac{\pi}{2}+n\pi, \tag*{(48)}

whereas for TOVFB, ζ(Rout)=0\zeta(R_{\mathrm{out}})=0 requires

θ(Rout)=nπ.(49)\theta(R_{\mathrm{out}})=n\pi. \tag*{(49)}

The integer nn labels the phase branches and is related to the number of interior nodes of ζ\zeta. In particular, the fundamental mode has no interior nodes and, with our choice of initial phase, satisfies θ(Rout)=π/2\theta(R_{\mathrm{out}})=\pi/2 for TOVFO and θ(Rout)=π\theta(R_{\mathrm{out}})=\pi for TOVFB.

For each background solution, we identify the critical radius RcR_{\mathrm{c}} by integrating the zero-frequency Prüfer equation and locating the first radius at which the corresponding fundamental-mode phase condition is satisfied. For TOVFO, this occurs at θ=π/2\theta=\pi/2, whereas for TOVFB it occurs at θ=π\theta=\pi. These radii define the marginal-stability thresholds, at which ω02=0\omega_0^2=0, separating the stable and unstable branches of the fundamental radial mode.

For TOVFO, this occurs at θ=π/2\theta=\pi/2, whereas for TOVFB it occurs at θ=π\theta=\pi. These radii define the marginal-stability thresholds, at which ω02=0\omega_0^2=0, separating the stable and unstable branches of the fundamental radial mode.

Table I shows coefficients of power-law fits X=CXμpXX=C_X\mu^{p_X} for the firehole radius RR, plateau mass MM, and critical radius RcR_{\mathrm{c}} for open (O) and boxed (B) boundary conditions. Fits use 10≤μ≤10710\leq\mu\leq10^7. The predicted exponent from Eq. (20) is p∗(q)=(1+2q)/5p_*(q)=(1+2q)/5.

The numerical results reveal that the critical radius shares approximately the same scaling with the central mass parameter as the equilibrium firehole radius and mass. As shown in Table I, the fitted exponents for RcR_{\mathrm{c}}, for both boundary conditions, are in close agreement with the prediction p∗(q)=(1+2q)/5p_*(q)=(1+2q)/5 of Eq. (20). Thus, the onset of radial instability is governed by a characteristic length scale that follows the equilibrium scaling. The agreement extends over the range 10≤μ≤10710\leq\mu\leq10^7 and for all the values of qq examined. The small, qq-dependent departures of the fitted exponents from p∗p_* reflect the slow approach of R/R∗R/R_* to its limiting value, which, as shown in Sec. III C, is controlled by β=μ(4−2q)/5\beta=\mu^{(4-2q)/5} and is therefore slower for larger qq.

A more interesting result emerges from the ratio of the critical radius to the firehole radius. For TOVFO, the critical radius is systematically smaller than the firehole radius, whereas for TOVFB the two radii nearly coincide. Over the range examined, 0.83≲RcO/R≲0.8980.83\lesssim R_{\mathrm{c}}^{\mathrm{O}}/R\lesssim0.898, with the lower end attained at the smallest β\beta (q=1q=1, μ=10\mu=10), while RB/RR_{\mathrm{B}}/R differs from unity by less than 10−210^{-2} (Figure 9). These results indicate that the location of the marginal-stability boundary depends sensitively on the outer boundary condition. In particular, confining the radial displacement at the outer boundary allows the configuration to remain stable up to a radius very close to the characteristic firehole radius RR.

Figure 9. Critical radii for 0≤q≤10 \leq q \leq1 and 10≤μ≤10710 \leq\mu\leq10^7, plotted against the scale-invariant parameter β=μ/R∗\beta= \mu/R_* of Eq. (63). Top: open boundary, RcO/RR_c^{\mathrm{O}}/R; the dashed line marks the large-β\beta limit 0.898. Bottom: boxed boundary, RcB/R−1R_c^{\mathrm{B}}/R - 1; the dashed line is 0.009/β0.009/\beta. Results for different qq fall on a single curve.

Fundamental zero-mode eigenfunctions

To examine the distinction between the two boundary conditions, we compare the zero-frequency solutions evaluated at their respective critical radii, RcOR_{\mathrm{c}}^{\mathrm{O}} and RcBR_{\mathrm{c}}^{\mathrm{B}}, for the same TOVF background. Figure 6 shows the displacement variable ζ\zeta, normalized by its maximum absolute value. The two zero modes are obtained from the same regular solution of the zero-frequency pulsation equations, truncated at their respective critical radii. Their agreement over the common radial interval is therefore expected; any small differences in the plotted curves arise from the independent amplitude normalizations and numerical accuracy. For TOVFO, the solution terminates at η(RcO)=0\eta(R_{\mathrm{c}}^{\mathrm{O}})=0, whereas for TOVFB it terminates at ζ(RcB)=0\zeta(R_{\mathrm{c}}^{\mathrm{B}})=0.

Normalized radial displacement for the fundamental zero modes

Figure 6. Normalized radial displacement ζ/max⁡∣ζ∣\zeta/\max|\zeta| for the fundamental zero modes of TOVFO and TOVFB, computed on the same equilibrium background and terminated at their respective critical radii. The dotted vertical line marks r=Rr=R.

The momentum-like variable η=Pζ′\eta=P\zeta' is shown separately for each boundary condition in Figure 7. Independent normalization by max⁡∣η∣\max|\eta| is used to resolve the shapes. In the open problem, η\eta reaches zero at RcOR_{\mathrm{c}}^{\mathrm{O}}. In the boxed problem, the vanishing of ζ\zeta at RcBR_c^{\mathrm{B}} is instead accompanied by a nonzero η\eta. Its rapid variation near the endpoint should not be interpreted as a discontinuity or a distinct inner eigenfunction.

Momentum-like variable for the fundamental zero modes

Figure 7. Momentum-like variable η=Pζ′\eta=P\zeta' for the fundamental zero modes: TOVFO (upper panel) and TOVFB (lower panel). Each curve is normalized by its own max⁡∣η∣\max|\eta| to display its radial structure; the relative magnitudes between panels are therefore not shown. The dotted vertical line marks r=Rr=R.

qqp∗p_\astCRC_RpRp_RCMC_MpMp_MCcOC_c^{\mathrm{O}}pcOp_c^{\mathrm{O}}CcBC_c^{\mathrm{B}}pcBp_c^{\mathrm{B}}
00.2001.31100.19590.61370.20111.14330.19821.31240.1958
0.250.3001.33060.29480.61380.30101.15280.29751.33240.2946
0.50.4001.35740.39330.61480.40071.16650.39661.36010.3931
0.750.5001.39470.49130.61730.50021.18690.49531.39920.4911
10.6001.44910.58880.62290.59921.21790.59341.45570.5884

Table 1. Power-law fits X=CXμpXX = C_X\mu^{p_X} for the firehole radius RR, plateau mass MM, and critical radius RcR_c for open (O) and boxed (B) boundary conditions. Fits use 10≤μ≤10710 \le\mu\le10^7, sampled at logarithmically spaced values. The predicted exponent is p∗(q)=(1+2q)/5p_\ast(q) = (1 + 2q)/5.

At zero frequency, Eqs. (24)–(25) imply the identity

(ζη)′=η2P−Qζ2.(50)(\zeta\eta)' = \frac{\eta^2}{P} - Q\zeta^2. \tag*{(50)}

Consequently, the cumulative variational numerator measured from the inner integration radius is

N(r)=∫R0rds[η2P−Qζ2]=ζ(r)η(r)−ζ(R0)η(R0).(51)\begin{aligned} \mathcal{N}(r) &= \int_{R_0}^{r} \mathrm{d}s\left[\frac{\eta^2}{P} - Q\zeta^2\right] \\ &= \zeta(r)\eta(r) - \zeta(R_0)\eta(R_0). \tag*{(51)} \end{aligned}

The endpoint term vanishes as R0→0R_0 \to0 for the regular inner branch. The two-panel comparison in Figure 8 illustrates how N(r)\mathcal{N}(r) accumulates in the respective zero modes. Its value vanishes at the critical radius for either condition, because the boundary product ζη\zeta\eta vanishes. The sign of this cumulative quantity at an intermediate radius must not be construed as the stability of an independently truncated configuration unless its boundary condition is also satisfied there.

Cumulative zero-mode variational numerator for TOVFO and TOVFB

Figure 8. Cumulative zero-mode variational numerator N(r)\mathcal{N}(r) for TOVFO (upper panel) and TOVFB (lower panel), independently normalized by max⁡∣N∣\max|\mathcal{N}|. The identity in Eq. (51) provides a direct consistency check. The dotted vertical line marks r=Rr=R.

Variational interpretation of the stability thresholds

The different critical radii obtained for TOVFO and TOVFB can be understood in terms of the variational properties of the radial pulsation operator. Multiplying Eq. (23) by ζ\zeta and integrating by parts yields

ω2=∫0Rout[P(ζ′)2−Qζ2]dr∫0RoutWζ2dr,(52)\omega^2 = \frac{\int_{0}^{R_{\mathrm{out}}}\left[P(\zeta')^2-Q\zeta^2\right]\mathrm{d}r}{\int_{0}^{R_{\mathrm{out}}}W\zeta^2\mathrm{d}r}, \tag*{(52)}

where the boundary term vanishes for both problems, and the lower limit is understood as R0→0R_0 \to0 along the regular branch, as in Eq. (51). At the inner endpoint, the regular solution satisfies ζη∼r5→0\zeta\eta\sim r^5 \to0. At the outer endpoint, the boundary term vanishes because η(Rout)=\eta(R_{\mathrm{out}}) = 00 for TOVFO and ζ(Rout)=0\zeta(R_{\mathrm{out}})=0 for TOVFB. Since P>0P>0 and W>0W>0, Eq. (52) provides the Rayleigh quotient associated with the radial pulsation problem.

The fundamental eigenvalue is characterized by the minimum of this quotient over the admissible radial perturbations. The two boundary conditions differ in their admissible function spaces. For TOVFO, η(Rout)=0\eta(R_{\mathrm{out}})=0 is the natural boundary condition associated with the variational problem, whereas TOVFB imposes the additional restriction ζ(Rout)=0\zeta(R_{\mathrm{out}})=0. Consequently, the admissible variations for TOVFB form a subset of those for TOVFO. For the same equilibrium background and outer radius RoutR_{\mathrm{out}}, the variational principle gives

ω0,O2(Rout)≤ω0,B2(Rout).(53)\omega_{0,\mathrm{O}}^2(R_{\mathrm{out}})\leq\omega_{0,\mathrm{B}}^2(R_{\mathrm{out}}). \tag*{(53)}

Here the subscripts O and B denote the open and boxed problems, respectively. This ordering assumes the same regular inner boundary condition and the corresponding self-adjoint realization of the pulsation operator.

Equation (53) establishes that the rigid outer boundary cannot lower the fundamental squared frequency. Thus, confining the fluid displacement has a stabilizing effect. In particular, if the fundamental eigenvalue of TOVFO is positive at a given outer radius, the corresponding boxed configuration must also be stable against radial perturbations.

The ordering also has a direct interpretation in terms of the zero-frequency solution. As discussed in Sec. III A, the regular zero mode is common to both boundary-value problems. Starting at the origin, ζ\zeta and η\eta are positive.

The first zero of η\eta, corresponding to an extremum of ζ\zeta, determines RcOR_c^{\mathrm{O}}. Continuing the same solution outward, η\eta becomes negative and ζ\zeta decreases until its first zero at RcBR_c^{\mathrm{B}}. Therefore,

RcO<RcB,(54)R_c^{\mathrm{O}}<R_c^{\mathrm{B}}, \tag*{(54)}

in agreement with our numerical results.

The ordering in Eq. (53) holds at fixed RoutR_{\mathrm{out}} for the same equilibrium background. It does not require the two eigenfunctions to have the same radial profile. At zero frequency, however, their critical solutions are different truncations of the same regular solution, which establishes the ordering of the critical radii.

The cumulative variational numerator introduced in Eq. (51) provides another way to understand this behavior. In the limit R0→0R_0\to0, the zero-frequency identity gives

N(r)=ζ(r)η(r).(55)\mathcal{N}(r)=\zeta(r)\eta(r). \tag*{(55)}

For the fundamental zero mode, ζ\zeta remains positive throughout the interior. Hence, N\mathcal{N} is positive up to RcOR_c^{\mathrm{O}}, vanishes at the first zero of η\eta, becomes negative between the two critical radii, and returns to zero when ζ\zeta vanishes at RcBR_c^{\mathrm{B}}. This behavior is illustrated in Figure 8.

The negative cumulative numerator in the interval RcO<r<RcBR_c^{\mathrm{O}}<r<R_c^{\mathrm{B}} does not imply that the boxed configuration is unstable. Indeed, at Rout=RcBR_{\mathrm{out}}=R_c^{\mathrm{B}}, the same zero-mode solution satisfies the rigid-wall boundary condition and has ω02=0\omega_0^2=0. Stability is determined by the eigenvalue of the complete boundary-value problem, rather than by the sign of the cumulative integral at an intermediate radius.

Zero modes as neighboring equilibria

The near-coincidence RcB≃RR_c^{\mathrm{B}}\simeq R has a simple origin: the regular zero-frequency solution is the displacement that connects neighboring members of the equilibrium family of Sec. II.

Consider a one-parameter family of TOVF equilibria, each a solution of the TOV Eqs. (5) and (6) with the same constant KK, and let δ\delta denote the derivative with respect to the family parameter at fixed rr. Because every member solves the TOV equations, the variations δm\delta m, δρ\delta\rho, δp\delta p, δν\delta\nu satisfy their linearization. To compare with the pulsation problem, we describe the passage from one member to its neighbor as a radial displacement of the fluid, δr=ξ(r)\delta r=\xi(r), which is the displacement of Eq. (22) at ω=0\omega=0. We define ξ\xi through the change in the enclosed mass at fixed rr,

δm=−4πr2(ρ+p)ξ,(56)\delta m=-4\pi r^2(\rho+p)\xi, \tag*{(56)}

or equivalently δλ=−(ν′+λ′)ξ\delta\lambda=-(\nu'+\lambda')\xi. In the derivation of Eq. (23), this relation follows from the trtr component of the perturbed Einstein equations, δtδλ=−(ν′+λ′)δtξ\delta_t\delta\lambda=-(\nu'+\lambda')\delta_t\xi, integrated in time from the unperturbed state. Combined with δm=re−2λδλ\delta m = r e^{-2\lambda}\delta\lambda, which follows from e−2λ=1−2m/re^{-2\lambda}=1-2m/r, and the background identity ν′+λ′=4πr(ρ+p)e2λ\nu' + \lambda' = 4\pi r(\rho+p)e^{2\lambda}, it gives Eq. (56), which therefore holds for any equation of state: the change in mm at fixed rr is the mass-energy carried across rr by the displaced fluid. Between static equilibria there is no time evolution, and Eq. (56) serves as the definition of ξ\xi, chosen to coincide with the ω→0\omega\to0 limit of the displacement in Eq. (23).

Varying Eq. (2) gives δm′=4πr2δρ\delta m' = 4\pi r^2\delta\rho. Together with Eq. (56) and p′=−(ρ+p)ν′p'=-(\rho+p)\nu' from Eqs. (3) and (4), this yields the Lagrangian density perturbation

Δρ≡δρ+ρ′ξ=−(ρ+p)[1r2(r2ξ)′−ν′ξ]=−(ρ+p)eνr2ζ′,(57)\begin{aligned} \Delta\rho\equiv\delta\rho+\rho'\xi=-(\rho+p)\left[\frac{1}{r^2}(r^2\xi)'-\nu'\xi\right] \\ &=-(\rho+p)\frac{e^\nu}{r^2}\zeta', \tag*{(57)} \end{aligned}

where the last step uses Eq. (22). The variations of Eqs. (3) and (4) hold automatically, since each member of the family is an equilibrium. Because every member obeys p=ρ/3p=\rho/3, the Lagrangian perturbations satisfy Δp=Δρ/3=γpΔρ/(ρ+p)\Delta p=\Delta\rho/3=\gamma p\Delta\rho/(\rho+p) with the same γ=4/3\gamma=4/3 that enters PP. This is the adiabatic relation assumed in the pulsation equation. The correspondence between neighboring equilibria and zero modes therefore requires the adiabatic index of the perturbations to equal (ρ+p)p−1dp/dρ(\rho+p)p^{-1}dp/d\rho evaluated along the equilibrium family, i.e., γ=Γ\gamma=\Gamma. For thermal radiation this holds exactly, since adiabatic compression of a photon gas preserves p=ρ/3p=\rho/3. With this relation, Δp=−γpeνζ′/r2\Delta p=-\gamma p e^\nu\zeta'/r^2, which is the relation between Δp\Delta p and ζ\zeta used in deriving Eq. (23), and the linearized hydrostatic equilibrium, Eq. (3), reduces to Eq. (23) at ω=0\omega=0. Finally, using ρ+p=e−4ν/(3π)\rho+p=e^{-4\nu}/(3\pi), which follows from 4πρe4ν=14\pi\rho e^{4\nu}=1 and p=ρ/3p=\rho/3, Eqs. (22) and (56) give

ζδ≡r2e−νξ=−34e3νδm,(58)\zeta_\delta\equiv r^2e^{-\nu}\xi=-\frac{3}{4}e^{3\nu}\delta m, \tag*{(58)}

which solves Eq. (23) with ω=0\omega=0. For a general Γ\Gamma-law fluid, Eq. (4) integrates to 4πρeΓν/(Γ−1)=K4\pi\rho e^{\Gamma\nu/(\Gamma-1)}=K, and the same steps give ζδ=−(ΓK)−1eν/(Γ−1)δm\zeta_\delta=-(\Gamma K)^{-1}e^{\nu/(\Gamma-1)}\delta m, which reduces to Eq. (58) for Γ=4/3\Gamma=4/3 and K=1K=1. This is the general-relativistic form of the classic correspondence between zero-frequency modes and neighboring equilibria [15]. In the same normalization (with the factor 3π3\pi absorbed into PP as above), the momentum-like variable is the Lagrangian pressure perturbation, ηδ=Pζδ′=−3πeλ+2νΔp\eta_\delta=P\zeta'_\delta=-3\pi e^{\lambda+2\nu}\Delta p, with

Δp=112πr2(3ν′δm−4r2e−4νδν).(59)\Delta p=\frac{1}{12\pi r^2}(3\nu'\delta m-4r^2e^{-4\nu}\delta\nu). \tag*{(59)}

The inner expansion, Eq. (8), identifies which family derivative is regular. Varying the inner lapse normalization at fixed central mass gives ∂Am∣μ=−2r5/(5A3)\partial_A m|_\mu=-2r^5/(5A^3) near the origin, so ζδ∝e3νr5∝r7/2\zeta_\delta\propto e^{3\nu}r^5\propto r^{7/2}, i.e. ξ∝r\xi\propto r: the regular branch. Varying the central mass at fixed AA gives ∂μm∣A=−1\partial_\mu m|_A=-1, so ζδ∝r−3/2\zeta_\delta\propto r^{-3/2}, i.e. ξ∝r−4\xi\propto r^{-4}: the discarded singular branch. The singular solution therefore

corresponds to a change of the central point mass, which a regular fluid perturbation cannot produce. Since the regular zero-frequency solution is unique up to normalization,

ζ0(r)∝e3ν∂m∂A∣μ,K.(60)\zeta_0(r)\propto e^{3\nu}\left.\frac{\partial m}{\partial A}\right|_{\mu,K}. \tag*{(60)}

We verified Eq. (60) numerically: the directly integrated zero-frequency solution and the finite-difference family derivative agree to better than 10−710^{-7} in relative amplitude over the full radial range, for all backgrounds examined. Equation (60) gives both thresholds a turning-point interpretation. For TOVFB, ζ(Rout)=0\zeta(R_{\mathrm{out}})=0 (equivalently ξ(Rout)=0\xi(R_{\mathrm{out}})=0) is equivalent to

∂m(Rout)∂A∣μ,K=0,(61)\left.\frac{\partial m(R_{\mathrm{out}})}{\partial A}\right|_{\mu,K}=0, \tag*{(61)}

so RcBR_c^{\mathrm{B}} is the radius at which the mass enclosed by the wall is stationary along the family of equilibria with fixed central mass and fixed KK. This is the analog, for a confined configuration, of the familiar criterion that the fundamental radial mode of a star changes stability at an extremum of the mass along the equilibrium sequence [15, 16, 17]. For TOVFO, η(Rout)=0\eta(R_{\mathrm{out}})=0 is equivalent to Δp(Rout)=0\Delta p(R_{\mathrm{out}})=0 in Eq. (59). The Lagrangian pressure of the boundary fluid element is stationary along the same family.

Scaling symmetry and the origin of RcB≃RR_c^{\mathrm{B}}\simeq R

At fixed KK, the TOV equations in the form of Eqs. (5) and (6) are invariant under

r→sr,m→sm,y→sy,(62)r\to sr,\qquad m\to sm,\qquad y\to sy, \tag*{(62)}

which maps μ→sμ\mu\to s\mu, A→s2AA\to s^2A, and R∗→sR∗R_*\to sR_*. Up to this rescaling, the singular solutions form a one-parameter family labeled by the scale-invariant ratio

β≡μR∗=(μ45A2)1/5=μ(4−2q)/5,(63)\beta\equiv\frac{\mu}{R_*}=\left(\frac{\mu^4}{5A^2}\right)^{1/5}=\mu^{(4-2q)/5}, \tag*{(63)}

and the mass function takes the form

m(r;μ,A)=R∗F(x;β),x=r/R∗.(64)m(r;\mu,A)=R_*F(x;\beta),\qquad x=r/R_*. \tag*{(64)}

Every dimensionless ratio, including R/R∗R/R_*, M/R∗M/R_*, and Rc/RR_c/R, depends on μ\mu and qq only through β\beta; the exponent qq merely selects the path β(μ)\beta(\mu). Using ∂ln⁡R∗/∂ln⁡A∣μ=2/5\partial\ln R_*/\partial\ln A|_\mu=2/5 and ∂ln⁡β/∂ln⁡A∣μ=−2/5\partial\ln\beta/\partial\ln A|_\mu=-2/5,

A∂m∂A∣μ=−25[r2(mr)′+R∗β∂F∂β].(65)A\left.\frac{\partial m}{\partial A}\right|_\mu=-\frac{2}{5}\left[r^2\left(\frac{m}{r}\right)'+R_*\beta\frac{\partial F}{\partial\beta}\right]. \tag*{(65)}

The approach of R/R∗R/R_* and M/R∗M/R_* to constant values found in Sec. II expresses that, outside the interior core, FF becomes independent of β\beta as β→∞\beta\to\infty. In that limit, the zero of Eq. (65) is the zero of (m/r)′(m/r)', which is the definition of the firehole radius RR. Hence

lim⁡β→∞RcBR=1.(66)\lim_{\beta\to\infty}\frac{R_c^{\mathrm{B}}}{R}=1. \tag*{(66)}

The relation RcB≃RR_c^{\mathrm{B}} \simeq R is therefore not a numerical coincidence: the rigid-wall zero mode vanishes where the equilibrium compactness m/rm/r is extremal. Figure 9 confirms both the collapse onto a single function of β\beta and the approach to this limit, with RcB/R−1≃0.009/βR_c^{\mathrm{B}}/R - 1 \simeq0.009/\beta for large β\beta. No comparable closed form follows for the open threshold; numerically, RcO/R→0.898R_c^{\mathrm{O}}/R \to0.898.

The 1/β1/\beta approach of RcB/RR_c^{\mathrm{B}}/R is faster than the convergence of the equilibrium scales themselves: numerically, R/R∗−R^∞R/R_* - \widehat{R}_\infty decreases only roughly as β−1/2\beta^{-1/2}. The argument leading to Eq. (66) establishes the limit but not this rate. The faster convergence is consistent with the leading large-β\beta correction to FF acting approximately as a rescaling of xx, F≃(1+ϵ)F∞(x/(1+ϵ))F \simeq(1+\epsilon)F_\infty\left(x/(1+\epsilon)\right) with ϵ(β)→0\epsilon(\beta) \to0. Such a deformation shifts RR and RcBR_c^{\mathrm{B}} together: its contribution to β∂F/∂β\beta\partial F/\partial\beta is proportional to F−xFxF-xF_x, which vanishes at x=R/R∗x=R/R_*, so it does not move the zero of Eq. (65) relative to RR.

Confinement bounds and the unbounded atmosphere

The rigid wall is not merely one admissible choice of outer boundary; it is the most stabilizing one. Consider any separated, self-adjoint condition at RoutR_{\mathrm{out}}, ζ(Rout)cos⁡α−κ−1η(Rout)sin⁡α=0\zeta(R_{\mathrm{out}})\cos\alpha-\kappa^{-1}\eta(R_{\mathrm{out}})\sin\alpha=0, which in Prüfer form reads θ(Rout)=α+nπ\theta(R_{\mathrm{out}})=\alpha+n\pi with α∈(0,π]\alpha\in(0,\pi]; TOVFO and TOVFB correspond to α=π/2\alpha=\pi/2 and α=π\alpha=\pi. Because θ(Rout;ω2)\theta(R_{\mathrm{out}};\omega^2) increases monotonically with ω2\omega^2, the fundamental eigenvalue increases monotonically with α\alpha. Its maximum, at α=π\alpha=\pi, is the boxed eigenvalue. Thus, for every such boundary condition,

ω0,α2(Rout)≤ω0,B2(Rout).(67)\omega_{0,\alpha}^2(R_{\mathrm{out}}) \leq\omega_{0,\mathrm{B}}^2(R_{\mathrm{out}}). \tag*{(67)}

Moreover, ω0,B2\omega_{0,\mathrm{B}}^2 decreases monotonically with RoutR_{\mathrm{out}}: the boxed eigenfunction on [0,Rout][0,R_{\mathrm{out}}], extended by zero, is an admissible trial function in Eq. (52) on any larger domain. The same construction applies to the complete atmosphere. For Rout>RcBR_{\mathrm{out}}>R_c^{\mathrm{B}}, the boxed fundamental mode, extended by zero to r>Routr>R_{\mathrm{out}}, is a compactly supported perturbation with regular inner behavior and negative Rayleigh quotient. The radial pulsation operator on [0,∞)[0,\infty) therefore has negative spectrum for any boundary condition at infinity whose form domain contains compactly supported perturbations. Because the trial function vanishes outside RoutR_{\mathrm{out}}, this conclusion is insensitive to the self-similar, non-asymptotically flat structure of the outer atmosphere.

Taken together, these results imply that a TOVF whose fluid extends beyond RcB≃RR_c^{\mathrm{B}}\simeq R is radially unstable, whether it is truncated by any separated, self-adjoint outer boundary condition or retains its complete atmosphere. Confinement can stabilize the configuration only if it acts within the narrow interval RcO<Rout<RcBR_c^{\mathrm{O}}<R_{\mathrm{out}}<R_c^{\mathrm{B}}, i.e., 0.90 R≲Rout≲R0.90\,R\lesssim R_{\mathrm{out}}\lesssim R for large β\beta, or at smaller radii. Frequency-dependent boundary conditions, such as coupling to a dynamical exterior medium, lie outside this argument.

Conclusions

We have investigated the stability against radial perturbations of TOVFs with a radiation equation of state, p=ρ/3p=\rho/3. These configurations consist of a negative point-mass singularity surrounded by positive-density fluid. Their mass profiles exhibit a rapid transition from the negative-mass interior to a Schwarzschild-like plateau, followed by an extended atmosphere approaching a self-similar asymptotic solution. Since the pressure does not vanish at any finite radius, the radial stability problem requires the specification of an outer boundary condition.

We have considered two boundary-value problems: TOVFO, for which the Lagrangian pressure perturbation vanishes at the outer boundary, η(Rout)=0\eta(R_{\mathrm{out}})=0, and TOVFB, for which the radial displacement vanishes, ζ(Rout)=0\zeta(R_{\rm out}) = 0. Using the Prüfer phase method, we identify the fundamental radial mode and determine the critical radius RcR_c at which its squared frequency vanishes. For both boundary conditions, the fundamental mode is stable for Rout<RcR_{\rm out}<R_c and unstable for Rout>RcR_{\rm out}>R_c in the range examined. For the boxed condition, this holds for every RoutR_{\rm out}, since the fundamental boxed eigenvalue decreases monotonically with RoutR_{\rm out}.

The numerical results reveal that the critical radii approximately follow the same scaling with the central negative mass parameter μ\mu as the equilibrium firehole radius RR and plateau mass MM. For families parametrized by the inner lapse normalization A=μq/5A=\mu^q/\sqrt{5}, we find

R, M, Rc∝μ(1+2q)/5,(68)R,\ M,\ R_c \propto\mu^{(1+2q)/5}, \tag*{(68)}

in agreement with the equilibrium scaling discussed in Sec. II. The dependence on qq reflects the choice of inner normalization used to construct a family of solutions, rather than a unique mass–radius relation determined by the TOV equations. A scaling symmetry of the TOV equations reduces the singular solutions, up to an overall scale, to a one-parameter family labeled by β=μ/R∗\beta=\mu/R_\ast; all dimensionless ratios, including Rc/RR_c/R, depend only on β\beta.

A central result of this study is that the two boundary conditions lead to systematically different critical radii. For TOVFO, the marginal-stability radius lies inside the characteristic firehole radius, RcO<RR_c^{\rm O}<R, approaching RcO≃0.898 RR_c^{\rm O}\simeq0.898\,R for large β\beta. In contrast, for TOVFB, the critical radius nearly coincides with it, RcB≃RR_c^{\rm B}\simeq R, over the parameter range examined. Thus, confinement of the radial displacement allows the configuration to remain stable over a larger radial domain, extending nearly to the firehole radius.

The different stability thresholds can be understood through the variational properties of the radial pulsation operator. For the same equilibrium background and outer radius, the rigid-wall condition restricts the space of admissible perturbations relative to the open condition. Consequently, the fundamental squared frequencies satisfy

ω0,O2(Rout)≤ω0,B2(Rout),(69)\omega_{0,\mathrm{O}}^2(R_{\rm out}) \leq\omega_{0,\mathrm{B}}^2(R_{\rm out}), \tag*{(69)}

under the corresponding self-adjoint boundary conditions. This establishes the stabilizing effect of confinement and is consistent with the ordering RcO<RcBR_c^{\rm O}<R_c^{\rm B}.

The zero-frequency eigenfunctions provide a complementary interpretation. Both boundary-value problems select different truncations of the same regular zero-mode solution. The first zero of the momentum-like variable η=Pζ′\eta=P\zeta' determines RcOR_c^{\rm O}, while the subsequent first zero of ζ\zeta determines RcBR_c^{\rm B}. Furthermore, the zero-frequency identity N(r)=ζ(r)η(r)\mathcal{N}(r)=\zeta(r)\eta(r), in the limit of a vanishing inner integration radius, relates the cumulative variational numerator directly to the eigenfunctions. Its behavior illustrates why the sign of a cumulative contribution at an intermediate radius cannot by itself determine the stability of a configuration with a different outer boundary condition.

This regular zero mode is the displacement that connects neighboring equilibria with the same central mass, generated by varying the inner lapse normalization AA. The boxed threshold is therefore a turning point: RcBR_c^{\rm B} is the radius at which the mass enclosed by the wall is stationary along this family. Combined with the scaling symmetry, this shows that the rigid-wall zero mode vanishes where the equilibrium quotient m/rm/r is extremal in the limit of large β\beta, which is the definition of the firehole radius. The near-coincidence RcB/R−1≃0.009/βR_c^{\rm B}/R-1\simeq0.009/\beta is thus explained, with RcB/R−1≃0.009/βR_c^{\rm B}/R-1\simeq0.009/\beta.

Our results demonstrate that the radial stability of TOVFs cannot be characterized independently of the boundary conditions imposed on their extended fluid atmosphere. In particular, the instability of an open configuration does not imply instability of the same background when confined by a rigid boundary placed in the interval RcO<Rout<RcBR_c^{\rm O}<R_{\rm out}<R_c^{\rm B}, that is, approximately between 0.9 R0.9\,R and RR. At the same time, the presence of a negative central singularity does not, by itself, determine the sign of the fundamental radial eigenvalue. Extending the analysis to physically specified, dynamical outer boundaries remains an open question.

An important implication of our results concerns the possibility of regarding TOVFs as isolated, stable compact objects. Among all separated, self-adjoint outer boundary conditions, the rigid wall yields the largest fundamental eigenvalue, and even under this condition the critical radius nearly coincides with the characteristic firehole radius, RcB≃RR_c^{\rm B}\simeq R. Extending the fluid domain beyond this radius introduces an unstable fundamental radial mode. Moreover, the boxed unstable mode, extended by zero, is a compactly supported perturbation of the complete, unbounded atmosphere with negative Rayleigh quotient. Since TOVFs possess an atmosphere extending indefinitely, these configurations are radially unstable as isolated objects, and stability requires an additional physical mechanism that confines the surrounding fluid on a scale comparable to the firehole radius.

Acknowledgments

This work was partially supported by NSF grants PHY-2411068 and PHY-2207780. The authors thank Fumika Suzuki, Wojciech Zurek, and Natalia Dinkus for helpful discussions. During the preparation of this manuscript, the authors used Claude (Anthropic), a large language model, to review the draft. The model independently verified the equilibrium and zero-frequency pulsation integrations, reproducing the numerical results reported here, including the fits in Table I, and suggested revisions to notation and presentation. The authors take full responsibility for all analytic results, numerical computations, and conclusions of this paper.

References

  1. [1]R. C. Tolman, Static solutions of Einstein’s field equations for spheres of fluid, Physical Review 55, 364 (1939).DOI
  2. [2]J. R. Oppenheimer and G. M. Volkoff, On massive neutron cores, Physical Review 55, 374 (1939).DOI
  3. [3]S. Chandrasekhar, The dynamical instability of gaseous masses approaching the Schwarzschild limit in general relativity, The Astrophysical Journal 140, 1342 (1964), erratum: Astrophys. J. 140, 1342 (1964).
  4. [4]S. Chandrasekhar, Dynamical instability of gaseous masses approaching the Schwarzschild limit in general relativity, Physical Review Letters 12, 114 (1964), erratum: Phys. Rev. Lett. 12, 437 (1964).
  5. [5]T. Hinderer, Tidal Love numbers of neutron stars, The Astrophysical Journal 677, 1216 (2008).DOI
  6. [6]K. Yagi and N. Yunes, I-Love-Q: Unexpected universal relations for neutron stars and quark stars, Science 341, 365 (2013).DOI
  7. [7]G. B. Cook, Initial data for numerical relativity, Living Reviews in Relativity 3, 5 (2000).
  8. [8]T. W. Baumgarte and S. L. Shapiro, Numerical relativity and compact binaries, Physics Reports 376, 41 (2003).DOI
  9. [9]M. D. Duez, Numerical relativity confronts compact neutron star binaries: A review and status report, Classical and Quantum Gravity 27, 114002 (2010), arXiv:0912.3529 [gr-qc].DOI
  10. [10]W. H. Zurek and D. N. Page, Black-hole thermodynamics and singular solutions of the Tolman–Oppenheimer–Volkoff equation, Physical Review D 29, 628 (1984).
  11. [11]G. ’t Hooft, The self-screening Hawking atmosphere: A different approach to quantum black hole microstates, Nuclear Physics B - Proceedings Supplements 68, 174 (1998), arXiv:gr-qc/9706058.DOI
  12. [12]C. Anastopoulos and N. Savvidou, Classification theorem and properties of singular solutions to the Tolman–Oppenheimer–Volkoff equation, Class. Quantum Grav. 38, 075024 (2021).DOI
  13. [13]R. C. Tolman, Relativity, Thermodynamics and Cosmology (Clarendon Press, Oxford, 1934).
  14. [14]K. D. Kokkotas and J. Ruoff, Radial oscillations of relativistic stars, A&A 366, 565 (2001).
  15. [15]B. K. Harrison, K. S. Thorne, M. Wakano, and J. A. Wheeler, Gravitation Theory and Gravitational Collapse (University of Chicago Press, Chicago, 1965).
  16. [16]J. M. Bardeen, K. S. Thorne, and D. W. Meltzer, A catalogue of methods for studying the normal modes of radial pulsation of general-relativistic stellar models, The Astrophysical Journal 145, 505 (1966).DOI
  17. [17]R. D. Sorkin, A stability criterion for many-parameter equilibrium families, The Astrophysical Journal 257, 847 (1982).DOI

Paper details

vlnc:W_gxfsj5cf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 2

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

  • Physics and Astronomy

Claude (Anthropic)

View statement
Looking for related papers…
Contents