Relativistic Cowling Approximation

• Relativistic Cowling approximation is a method that fixes the spacetime metric to isolate fluid perturbations in compact-star oscillation studies.
• It reduces the full Einstein-fluid problem to a fluid eigenvalue problem with real mode frequencies, enhancing numerical efficiency.
• The approximation is widely used in asteroseismology and EOS surveys, though it excludes gravitational-wave backreaction and spacetime modes.

The relativistic Cowling approximation is a fixed-spacetime approximation for relativistic stellar perturbations in which the equilibrium star is constructed in full general relativity, while the oscillation problem is solved after neglecting Eulerian perturbations of the metric, so that only the matter degrees of freedom are evolved on a curved but frozen background spacetime (Córsico et al., 2023). In compact-star asteroseismology this converts the full coupled Einstein-fluid problem into a fluid eigenvalue problem with real frequencies, no gravitational-wave backreaction, and no purely spacetime modes, while retaining the relativistic structure of the background star (Ranea-Sandoval et al., 2018).

1. Definition and conceptual scope

In its standard form, the approximation is expressed as

$\delta g_{\mu\nu}=0,$

or equivalently, in $3+1$ language, by setting perturbations of the lapse and shift to zero in the fluid problem (Counsell et al., 2024). The defining physical statement is not that gravity is ignored, but that the gravitational field is held fixed while the fluid oscillates.

A sharp distinction therefore exists between the full relativistic perturbation problem and its Cowling reduction. In the full problem, fluid and metric perturbations are coupled, the mode frequencies are generally complex, outgoing-wave conditions are required, and spacetime-led oscillations such as $w$-modes are present. In the relativistic Cowling approximation, the gravitational degrees of freedom are set to zero, only the fluid perturbations remain, the eigenfrequencies are real, and gravitational-wave damping is absent (Córsico et al., 2023).

Aspect	Full relativistic perturbations	Relativistic Cowling approximation
Dynamical variables	Fluid + metric perturbations	Fluid perturbations only
Frequency character	Complex quasinormal modes	Real stationary modes
Radiation sector	Outgoing-wave conditions, GW emission	No GW backreaction or damping
Spacetime modes	Present	Absent

This also clarifies the relation to the Newtonian Cowling approximation. In the Newtonian case one neglects perturbations of the Newtonian gravitational potential, whereas in the relativistic case one neglects perturbations of the metric itself (Córsico et al., 2023). The approximation is therefore fully relativistic in the background and reduced only at the perturbative level.

2. Relativistic background structure

The background models used with the relativistic Cowling approximation are ordinarily static and spherically symmetric, with line element

$$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$

or equivalent notations such as $e^\nu$ and $e^\lambda$ in Schwarzschild-like coordinates (Ranea-Sandoval et al., 2018). The equilibrium star is then obtained from the Tolman–Oppenheimer–Volkoff equations, or from direct Einstein-equation rearrangements in applications that emphasize the metric potentials explicitly (Córsico et al., 2023).

Within this framework the equilibrium variables are relativistic quantities. In the ultra-massive white-dwarf treatment, for example, $m(r)$ is the total enclosed relativistic mass, including rest mass, nuclear binding energy, internal energy, and gravity, while $\rho$ is the mass-energy density rather than a purely Newtonian rest-mass density (Córsico et al., 2023). This point is central: Cowling does not alter the relativistic equilibrium configuration, only the perturbative response of spacetime.

The same strategy extends to nonstandard matter sectors. Charged strange-star calculations use the Einstein–Maxwell background, with

$$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$

and then perturb the fluid and charge/current variables while still imposing $\delta g_{\mu\nu}=0$ (Arbañil et al., 2024). Anisotropic neutron-star models similarly retain a fully relativistic background with radial pressure $3+1$0, tangential pressure $3+1$1, and generalized hydrostatic balance

$3+1$2

where $3+1$3, before applying the Cowling reduction to the perturbations (Doneva et al., 2012).

This suggests that the relativistic Cowling approximation is best viewed as a perturbative restriction rather than a matter-model restriction: it can be combined with perfect fluids, anisotropic stresses, electric charge, elasticity, or modified-gravity backgrounds, provided the equilibrium spacetime is specified.

3. Fluid eigenvalue problem and boundary conditions

For polar non-radial oscillations, the retained dynamical variables are usually the radial and tangential components of the Lagrangian displacement. Many compact-star papers write these as $3+1$4 and $3+1$5, through a spherical-harmonic expansion of the displacement vector, while the white-dwarf formulation in (Córsico et al., 2023) uses dimensionless functions $3+1$6 and $3+1$7 constructed from the radial and horizontal displacements $3+1$8 and $3+1$9. In either notation, the metric perturbations are removed and the problem collapses to two coupled first-order ODEs.

A standard relativistic Cowling system for polar modes is

$w$0

$w$1

with $w$2 the mode frequency, $w$3 the energy density, $w$4 the pressure, and $w$5 the spherical-harmonic degree (Ranea-Sandoval et al., 2018). For composition $w$6-modes in mature neutron stars, an equivalent formulation can be written in terms of $w$7 and $w$8, with the buoyancy physics encoded by the difference between the perturbed adiabatic index $w$9 and the background index $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$0 (Counsell et al., 2024).

The eigenvalue problem is completed by boundary and regularity conditions. At the center, regularity requires

$$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$1

for some normalization constant $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$2 (Ranea-Sandoval et al., 2018). At the stellar surface, the Lagrangian pressure perturbation must vanish,

$$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$3

which selects the allowed eigenfrequencies (Ranea-Sandoval et al., 2018).

When the equation of state contains a sharp discontinuity, additional interface conditions are required. In hybrid-star calculations, the radial displacement $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$4 and the Lagrangian pressure perturbation are continuous across the transition radius, while the angular variable $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$5 is generally discontinuous (Ranea-Sandoval et al., 2018). In proto-neutron-star plus shock calculations, the outer boundary can instead be taken at the stalled shock, with vanishing radial displacement there, because the oscillation cavity extends from the center to the shock rather than to the stellar surface (Torres-Forné et al., 2017).

Numerically, this structure makes shooting methods natural. One integrates from the center outward, applies any jump conditions at phase interfaces, and iterates on the trial frequency until the surface or shock boundary condition is satisfied. In the hybrid-star code CFK, this is implemented with Runge–Kutta–Fehlberg integration plus Newton–Raphson and Ridders root finding (Ranea-Sandoval et al., 2018).

4. Accuracy, mode dependence, and known limitations

The approximation is not uniformly accurate across mode families. For compact stars above roughly $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$6, one study of hybrid stars reports that Cowling $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$7-mode frequencies differ from full GR by about $$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$8–$$ds^2=-e^{2\Phi(r)}dt^2+e^{2\Lambda(r)}dr^2+r^2d\Omega^2,$$9, $e^\nu$0-mode errors are typically $e^\nu$1, and $e^\nu$2-mode errors are typically $e^\nu$3, with the accuracy improving for larger compactness (Ranea-Sandoval et al., 2018). A conference summary of the same general framework states more broadly that the fluid-mode frequencies differ by less than about $e^\nu$4 from those obtained from the full linearized Einstein-fluid system (Ranea-Sandoval et al., 2018).

Full-GR benchmarks sharpen this picture. For hyperonic neutron stars, the quadrupolar $e^\nu$5-mode frequency in Cowling can be overestimated by up to $e^\nu$6, and the discrepancy decreases with increasing stellar mass (Pradhan et al., 2022). In proto-neutron stars, the overall frequency accuracy is of order $e^\nu$7, with $e^\nu$8-modes reproduced within $e^\nu$9, $e^\lambda$0-modes within $e^\lambda$1, and the early-time $e^\lambda$2-mode within $e^\lambda$3; in that setting the $e^\lambda$4- and $e^\lambda$5-mode frequencies are basically overestimated, while the $e^\lambda$6-mode frequencies are basically underestimated (Sotani et al., 2020).

The most important structural limitation is that Cowling removes the radiative sector. Because the metric is frozen, the eigenfrequencies are purely real, gravitational-wave damping times cannot be computed, and purely spacetime modes such as the $e^\lambda$7-modes are excluded (Ranea-Sandoval et al., 2018). This is why full GR is required for quasinormal-mode damping times, outgoing-wave boundary conditions, and any precision calibration of $e^\lambda$8-mode asteroseismic relations (Pradhan et al., 2022).

Mode dependence can also be highly nontrivial in elastic problems. In stars with an elastic crust, the relativistic Cowling approximation has virtually no impact on the frequencies or eigenfunctions of polar shear modes, with subpercent differences in the tabulated examples, but interface modes associated with jumps in the shear modulus shift by roughly $e^\lambda$9–$m(r)$0 and can even exhibit different nodal structure between Cowling and full GR (Krüger et al., 2024). Weak gravitational-wave damping is therefore not, by itself, a sufficient indicator of Cowling accuracy.

5. Principal astrophysical applications

The approximation is used most often when the main target is the fluid spectrum rather than gravitational-wave damping. In ultra-massive ZZ Ceti white dwarfs, relativistic Cowling calculations on fully evolutionary ONe-core white-dwarf models show that the Brunt–Väisälä and Lamb frequencies are larger in the relativistic case than in the Newtonian case, and that the $m(r)$1-mode periods are shorter, with relative differences of up to $m(r)$2 for the $m(r)$3 models in the ZZ Ceti instability strip; the conclusion is that general-relativistic effects on white dwarfs with masses larger than $m(r)$4 cannot be ignored in asteroseismology (Córsico et al., 2023).

In hybrid stars, the approximation is especially useful for surveying equations of state. The key result is that $m(r)$5-modes are present only for a sharp first-order hadron–quark transition and disappear when the transition is smoothed into a mixed phase. For sharp transitions, the computed $m(r)$6-mode frequencies lie in the range $m(r)$7–$m(r)$8, and frequencies between $m(r)$9 and $\rho$0 are proposed as evidence for a sharp hadron–quark phase transition and a pure quark core (Ranea-Sandoval et al., 2018).

In core-collapse and proto-neutron-star asteroseismology, relativistic Cowling mode calculations have been used to identify $\rho$1-, $\rho$2-, $\rho$3-, and hybrid modes in the evolving proto-neutron star and post-shock cavity, and to connect these mode tracks to features in simulated gravitational-wave spectrograms (Torres-Forné et al., 2017). For mature neutron stars with composition stratification, the same framework has been used to compute core and crust $\rho$4-modes and to estimate tidal-resonance phase shifts during binary inspiral, with the conclusion that individual resonances may be within reach of future detectors such as Cosmic Explorer and the Einstein Telescope, especially for stars with $\rho$5 (Counsell et al., 2024).

The framework also accommodates more specialized matter sectors. In charged strange stars, Cowling calculations on Einstein–Maxwell backgrounds show that the $\rho$6-mode is very sensitive to electric charge while the $\rho$7-mode is much less affected; for the maximum-mass models in the tabulated sequence, the $\rho$8-mode increases from $\rho$9 to $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$0 as the charge parameter $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$1 varies from $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$2 to $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$3, whereas the $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$4-mode decreases from $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$5 to $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$6 (Arbañil et al., 2024).

6. Variants, extensions, and common misconceptions

A common misconception is that the Cowling approximation is “nonrelativistic.” The literature shows the opposite: the background star is often treated in full GR, sometimes with TOV structure, sometimes with Einstein–Maxwell fields, sometimes with anisotropic stresses, and only the perturbative spacetime response is suppressed (Córsico et al., 2023). In that sense the approximation is a frozen-spacetime reduction, not a Newtonianization of the background.

A second misconception is that there is a single relativistic Cowling scheme. In practice the approximation has several extensions. For anisotropic neutron stars, one perturbs only the conservation law $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$7 on a fixed anisotropic background, with $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$8 for the quasilocal compactness variable entering the anisotropy law (Doneva et al., 2012). In TeVeS, the approximation is generalized by freezing not only the metric but also the vector and scalar field perturbations, so that only the axial fluid perturbation is retained; in that setting the toroidal-mode equation has the same form as in GR, although the moment of inertia depends strongly on the gravity theory (Sotani, 2010). In GR resistive MHD, the fixed-spacetime version of the full GRRMHD system is likewise described as a Cowling approximation, with fluid and electromagnetic variables evolved on the initial curved metric (Dionysopoulou et al., 2012).

A third misconception is that “Cowling” always refers to oscillation spectra. In static deformation problems the term often denotes neglect of the perturbation’s self-gravity in Newtonian theory. A reassessment of magnetically induced ellipticity found that including gravitational-potential perturbations yields ellipticities nearly twice as large as those obtained in the Cowling approximation for the standard background models considered (Yoshida, 2013). Fully relativistic calculations of maximum elastic quadrupoles go further still: they generalize the old Newtonian Cowling-based integral method by retaining metric perturbations, and they find relativistic suppressions relative to the Newtonian Cowling estimate of up to $$e^{2\Lambda}=\left(1-\frac{2m}{r}+\frac{q^2}{r^2}\right)^{-1},$$9 for crustal deformations, up to $\delta g_{\mu\nu}=0$0 for hybrid stars, and at most $\delta g_{\mu\nu}=0$1 for solid quark stars (Johnson-McDaniel et al., 2012). This suggests that the meaning of “Cowling” is problem-dependent, even though the unifying idea is always the freezing of gravitational degrees of freedom.

Taken together, these results define the present role of the relativistic Cowling approximation. It is a technically efficient and often physically incisive reduction of the relativistic perturbation problem, particularly for buoyancy modes, pressure modes, crustal shear modes, and large parameter surveys. Its strengths lie in mode classification, EOS sensitivity studies, and exploratory asteroseismology; its weaknesses lie in GW damping, spacetime-led modes, and precision calibration of fluid-mode frequencies.