Einstein-Aether Gravity

• Einstein-aether gravity is a Lorentz-violating extension of general relativity that couples the metric to a dynamic unit timelike vector field (the aether) to establish a preferred local rest frame.
• The theory propagates tensor, vector, and scalar modes with coupling-dependent speeds, influencing gravitational wave propagation and strong-field phenomena like black holes and compact binaries.
• It serves as a versatile framework encompassing cosmological reconstructions, post-Newtonian dynamics, and exact solutions, while current multimessenger tests tightly constrain its parameter space.

Einstein-aether gravity is a generally covariant Lorentz-violating modification of general relativity in which the metric $g_{ab}$ is coupled to a dynamical unit timelike vector field $u^a$, the “aether,” which selects a preferred local rest frame while preserving diffeomorphism invariance. In its standard two-derivative form the theory propagates tensor, vector, and scalar gravitational modes with coupling-dependent speeds, admits a hypersurface-orthogonal sector equivalent to the infrared limit of khronometric or extended Hořava gravity, and has been developed across cosmology, compact objects, post-Newtonian dynamics, and gravitational-wave phenomenology. Modern analyses show that multimessenger and preferred-frame tests strongly compress the viable parameter space, but they do not trivialize the theory: specialized sectors still exhibit distinctive causal structure, exact rotating black holes, modified wave propagation, and nontrivial cosmological effective-fluid behavior (0711.3822, 0801.1547, Streibert et al., 2024).

1. Covariant structure and coupling bases

In a standard convention, the dynamical fields are the spacetime metric and a unit timelike vector $u^a$, with the norm constraint enforced by a Lagrange multiplier. A common form of the action is

$$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$

with

$$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$

and

$$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$

Varying the action gives Einstein equations with an aether stress tensor, an aether field equation, and the unit constraint. In the weak-field limit the locally measured Newtonian constant differs from the bare $G$; one standard relation is $G_N=G/(1-c_{14}/2)$, with $c_{14}\equiv c_1+c_4$ (Vylet et al., 2023, 0711.3822).

A useful reorganization decomposes $\nabla_a u_b$ into expansion, shear, twist, and acceleration: $u^a$0 In that basis the aether Lagrangian is written as

$u^a$1

with

$u^a$2

This basis is technically convenient because mode speeds and phenomenological sectors often depend on these combinations rather than on the individual $u^a$3 (Vylet et al., 2023, Jacobson, 2013).

Basis	Definitions	Main use
Standard $u^a$4	$u^a$5	Covariant action and field equations
Kinematic basis	$u^a$6	Expansion, shear, twist, acceleration decomposition
Khronometric notation	$u^a$7	Hypersurface-orthogonal/Hořava limit

A distinct cosmological generalization replaces the linear kinetic scalar by a free function $u^a$8, where $u^a$9 is built from derivatives of the aether and, in FRW symmetry with $u^a$0, satisfies

$u^a$1

with $u^a$2 a constant combination of couplings. This $u^a$3 framework is not the standard linear Einstein-aether theory, but it has become important in reconstruction-based cosmology (Debnath, 2013, Ranjit et al., 2014).

2. Hypersurface orthogonality and the Hořava connection

The aether is hypersurface-orthogonal when it is normal to a foliation by constant values of a scalar $u^a$4, so that

$u^a$5

Equivalently, the twist vanishes. In this sector the independent local content of the aether is reduced, the spin-1 mode disappears, and the theory becomes the covariant “khronometric” or $u^a$6-theory sector (Jacobson, 2010).

A central structural result is that the infrared limit of the extended Hořava theory of Blas, Pujolàs, and Sibiryakov is precisely the hypersurface-orthogonal sector of Einstein-aether theory. In $u^a$7 form the aether kinematics decomposes as

$u^a$8

and the æther Lagrangian reduces to combinations of $u^a$9, $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$0, and $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$1. Matching to the BPS infrared Hořava action gives a direct parameter map; with a common convention $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$2, $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$3, and $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$4. As a consequence, any hypersurface-orthogonal Einstein-aether solution is also a solution of the infrared BPS theory (Jacobson, 2010).

A complementary limit clarifies the relation from the Einstein-aether side. Sending the twist coupling to infinity,

$$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$5

while keeping $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$6 fixed forces $$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$7 for regular solutions, thereby selecting hypersurface orthogonality. In that limit Einstein-aether reproduces low-energy Hořava gravity, the spin-1 mode decouples, and quantities such as PPN relations, rotating black-hole equations, and radiation rates can be inherited from the Einstein-aether formulation (Jacobson, 2013).

This equivalence is often summarized too strongly. The precise statement is that the hypersurface-orthogonal sector of Einstein-aether coincides with the infrared khronometric theory, not that generic Einstein-aether and Hořava gravity are identical. The distinction matters whenever twist or spin-1 dynamics is observationally relevant (Jacobson, 2010, Jacobson, 2013).

3. Linear modes, preferred-frame effects, and empirical bounds

Linearization about Minkowski spacetime yields three sectors: two spin-2 tensor modes, two spin-1 vector modes, and one spin-0 scalar mode. In standard notation one convenient set of speeds is

$$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$8

$$S=\int d^4x\,\sqrt{-g}\,\Big[\frac{1}{16\pi G}R+\mathcal L_{\rm ae}+\mathcal L_m\Big],$$9

$$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$0

Stability requires nonnegative squared speeds and positive kinetic energies; absence of gravitational Cherenkov losses requires the relevant mode speeds to be at least luminal to very high accuracy (Streibert et al., 2024, 0711.3822).

In the post-Newtonian regime all PPN parameters agree with general relativity except the preferred-frame parameters $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$1 and $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$2. Early reviews quote bounds $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$3 and $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$4, while recent summaries use $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$5 and $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$6 (0711.3822, Vylet et al., 2023). There are coupling submanifolds on which $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$7, and much of the viable parameter space is discussed in those tuned sectors (0801.1547).

Multimessenger observations have sharply changed the phenomenology. Because

$$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$8

in the kinematic basis, GW170817/GRB170817A implies $$\mathcal L_{\rm ae}=-M^{ab}{}_{cd}\,(\nabla_a u^c)(\nabla_b u^d)+\lambda(u^a u_a+1),$$9 at the $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$0 level. Recent neutron-star analyses therefore organize the viable space into regions with $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$1, $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$2 very small, and $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$3 essentially unconstrained except for positivity of the spin-1 energy (Vylet et al., 2023).

Propagation on inhomogeneous backgrounds does not re-open this freedom in any obvious way. Gravitational-wave lensing in Einstein-aether modifies the local tensor speed through inhomogeneities in the background aether, but the modification is common to both tensor polarizations and vanishes in the luminal limit $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$4; no lens-induced birefringence appears at principal order (Streibert et al., 2024). More recently, a direct Isaacson-based calculation of gravitational displacement memory found that superluminal scalar or vector aether waves produce “unprotected causal directions” with potentially unbounded memory growth, leading to a conjectured exclusion of the superluminal parameter space (Heisenberg et al., 14 May 2025). That exclusion is presented explicitly as a conjecture rather than as a completed observational bound.

4. Cosmological sectors and reconstruction programs

In homogeneous FRW symmetry the aether is aligned with the cosmological rest frame, and the theory can be written as modified Friedmann equations with an effective aether energy density and pressure. In the generalized $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$5 formulation one has

$$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$6

and

$$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$7

These relations permit reconstruction of $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$8 by imposing $$M^{ab}{}_{cd}=c_1 g^{ab}g_{cd}+c_2\delta^a_c\delta^b_d+c_3\delta^a_d\delta^b_c+c_4 u^a u^b g_{cd}.$$9 for a chosen dark-energy model (Debnath, 2013).

Several reconstruction programs have been explored. For holographic and agegraphic models, power-law scale factors $G$0 and phantom-like forms $G$1 yield explicit $G$2. In the reconstructions from ordinary HDE and NADE, the power-law branch gives quintessence-like behavior while the phantom branch gives $G$3. Entropy-corrected HDE and NADE permit phantom crossing for both branches. In that analysis, however, the squared sound speed is negative in all four correspondences, so the reconstructed Einstein-aether dark-energy sectors are classically unstable within the parameter ranges studied (Debnath, 2013).

A separate reconstruction program matched Einstein-aether to modified gravities $G$4, $G$5, $G$6, $G$7, and $G$8 on power-law FRW backgrounds. There the reconstructed models from $G$9, $G_N=G/(1-c_{14}/2)$0, and $G_N=G/(1-c_{14}/2)$1 were classically stable according to the background-level criterion $G_N=G/(1-c_{14}/2)$2, the $G_N=G/(1-c_{14}/2)$3 correspondence was unstable, and the $G_N=G/(1-c_{14}/2)$4 correspondence changed stability during evolution (Ranjit et al., 2014). These results are not contradictory in a strict sense; they indicate that stability is highly reconstruction-dependent.

Beyond reconstruction, exact FLRW analyses show that Einstein-aether cosmology is not generally reducible to a mere renormalization of Newton’s constant. For a fluid with $G_N=G/(1-c_{14}/2)$5, exact solutions exhibit exponential expansion or contraction for $G_N=G/(1-c_{14}/2)$6, bounce solutions for $G_N=G/(1-c_{14}/2)$7, and finite-time curvature singularities for $G_N=G/(1-c_{14}/2)$8. In vacuum with $G_N=G/(1-c_{14}/2)$9, the theory admits nonflat FLRW solutions sourced only by the aether, with a curvature singularity at finite time unless $c_{14}\equiv c_1+c_4$0 (Campista et al., 2018).

Additional phenomenological cosmology used the $c_{14}\equiv c_1+c_4$1 formalism with Granda–Oliveros and Chen–Jing dark-energy ansätze, as well as Tsallis, Rényi, and Sharma–Mittal holographic models. The resulting Einstein-aether effective fluid can mimic quintessence, phantom, or $c_{14}\equiv c_1+c_4$2CDM-like behavior depending on the background ansatz and parameters, and Om-diagnostic analyses find redshift-dependent deviations from $c_{14}\equiv c_1+c_4$3CDM rather than a constant $c_{14}\equiv c_1+c_4$4 (Pasqua et al., 2015, Rani et al., 2019).

5. Compact objects, post-Newtonian dynamics, and gravitational waves

Strong-field bodies acquire sensitivities that encode how their masses depend on motion relative to the aether. These sensitivities modify conservative dynamics, produce preferred-frame effects, and generate dipole radiation. Reviews emphasize that binary pulsars and compact objects are therefore among the sharpest probes of the theory, especially after $c_{14}\equiv c_1+c_4$5 and $c_{14}\equiv c_1+c_4$6 are tuned small (0801.1547).

For compact binaries, a recent post-Minkowskian and DIRE construction derived relaxed field equations and near-zone solutions through $c_{14}\equiv c_1+c_4$7 post-Newtonian order in the GW-speed-safe sector $c_{14}\equiv c_1+c_4$8. Because the unit constraint would otherwise introduce source terms linear in the perturbations, the formalism uses superpotential-based field redefinitions to recover flat-spacetime wave equations for the redefined fields, with quadratic-and-higher nonlinear sources. This provides the technical foundation for equations of motion and waveforms for orbiting compact bodies in Einstein-aether gravity (Taherasghari et al., 2023).

Neutron-star universal relations remain remarkably insensitive to the twist coupling. In viable regions of parameter space, the I–Love–Q relations are very close to general relativity, and in one viable region they coincide exactly with it. The dependence on $c_{14}\equiv c_1+c_4$9 enters only through the aether’s rotational $\nabla_a u_b$0-component and is suppressed by $\nabla_a u_b$1; with realistic $\nabla_a u_b$2, deviations in $\nabla_a u_b$3, $\nabla_a u_b$4, and $\nabla_a u_b$5 are $\nabla_a u_b$6, far below EOS scatter. Consequently, I–Love–Q relations cannot constrain $\nabla_a u_b$7 (Vylet et al., 2023).

Direct comparison with interferometric data has now been performed. The waveform model EA_IMRPhenomD_NRT modifies IMRPhenomD_NRTidalv2 by including Einstein-aether corrections, additional scalar and vector polarizations, and strong priors derived from stability, cosmology, solar-system tests, binary pulsars, and GW170817. Parameter estimation on GW170817 and GW190425 found that present LIGO/Virgo data do not improve existing constraints, largely because the leading inspiral deviations are dominated by already constrained dipole effects and are partly degenerate with source parameters such as the chirp mass (Schumacher et al., 2023).

6. Exact solutions, specialized sectors, and extensions

Exact and near-exact solution theory in Einstein-aether gravity has broadened significantly. Nonrotating black holes and stars were established early, but a major recent development is the identification of the Kerr metric itself as an exact vacuum solution in a phenomenologically viable corner of parameter space, namely a “minimal æ-theory” with $\nabla_a u_b$8 and only $\nabla_a u_b$9. In that corner the associated aether has vanishing expansion, the exterior geometry retains the standard Kerr horizons and ergoregion, and the solution contains a spacelike quasi-universal-horizon surface inside the Killing horizon. It also removes the causality-violating region of the maximal Kerr extension because the aether becomes non-real there (Franzin et al., 2023).

Wave propagation in highly symmetric sectors can differ qualitatively from general relativity even before observational constraints are imposed. In cylindrical symmetry the Einstein-aether analog of the Einstein–Rosen wave equation is

$u^a$00

so the cylindrical spin-2 wave propagates with speed $u^a$01. For the class of solutions constructed in that setting, both metric functions $u^a$02 and $u^a$03 are periodic in time in Einstein-aether theory, whereas in general relativity $u^a$04 is periodic but $u^a$05 is only semi-periodic and exhibits secular drift for pulses (Chan et al., 2023).

Several specialized formulations expose otherwise hidden structure. In the scalar Einstein-aether theory, the vector is restricted to a gradient form $u^a$06, the timelike nature is imposed by a Lagrange multiplier, and a direct coupling between the matter flux and the aether produces non-geodesic motion and a generalized Poisson equation in the Newtonian limit (Haghani et al., 2014). In the spinning-aether extension, the aether acquires an intrinsic spin-rotation tensor $u^a$07, with new couplings $u^a$08, $u^a$09, and $u^a$10; linearized analysis shows that the spin-0 and spin-1 speeds are modified while the spin-2 speed remains $u^a$11 (Kohler, 2021).

The minimal $u^a$12-only sector has acquired a particularly sharp reinterpretation. In the version without an off-shell norm constraint, the aether action is Hodge dual to a $u^a$13-form gauge theory with $u^a$14-form field strength, so the metric equations reduce on shell to those of general relativity with a cosmological constant determined by $u^a$15. In that dual picture, any divergence-free aether field is pure gauge; the paper proves this as a theorem and uses it to discard certain proposed minimal-aether solutions as physically trivial (Hajian, 17 Apr 2025).

Taken together, these specialized sectors show that Einstein-aether gravity is not a single phenomenological template but a family of closely related Lorentz-violating gravitational theories. Some corners are now tightly pinned near general relativity by $u^a$16 and preferred-frame bounds, while others remain valuable as exact laboratories for causal structure, gauge duality, universal horizons, and strong-field solution theory.