Kinetic Gravity Braiding in Scalar-Tensor Theory

• Kinetic gravity braiding is a scalar-tensor framework featuring inherent kinetic mixing in a cubic Horndeski sector that produces second-order field equations.
• It employs two free functions, K and G, whose forms—often shift-symmetric—govern attractor cosmologies, perturbation dynamics, and self-tuning in dark energy models.
• KGB modifies observable signatures by affecting scalar perturbations and gravitational clustering, which in turn influence matter power spectra and ISW correlations.

Kinetic gravity braiding (KGB) is the cubic Horndeski, or $L_2+L_3$, sector of scalar–tensor theory in which a scalar field with Lagrangian $K(\phi,X)-G(\phi,X)\Box\phi$ exhibits an essential kinetic mixing between scalar and metric degrees of freedom while still yielding second-order field equations (Bernardo et al., 2020). In the minimally coupled realization one usually writes

$$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$

with $G_4=M_{\rm Pl}^2/2$ and $G_5=0$; in other sign conventions, especially with signature $(+---)$, the same sector is written as $K(\phi,X)+G(\phi,X)\Box\phi$ and $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$ (Nouri-Zonoz et al., 6 Nov 2025, Pujolas et al., 2011). The term “braiding” refers to the fact that the $G\Box\phi$ operator mixes scalar and tensor kinetic structures in a way that cannot be removed by field redefinitions, and this mixing underlies applications ranging from dark-energy phenomenology and inflation to stealth black holes, wormholes, null-shell junction conditions, and exact lower-dimensional dilaton-gravity embeddings (Deffayet et al., 2010, Takahashi et al., 2018).

1. Covariant structure and defining subclasses

The defining covariant data of KGB are the two free functions $K$ and $K(\phi,X)-G(\phi,X)\Box\phi$0. In many contemporary cosmological and black-hole applications, KGB denotes the minimally coupled Horndeski $K(\phi,X)-G(\phi,X)\Box\phi$1 sector with constant $K(\phi,X)-G(\phi,X)\Box\phi$2 and vanishing $K(\phi,X)-G(\phi,X)\Box\phi$3, so that all departures from GR arise through $K(\phi,X)-G(\phi,X)\Box\phi$4 and $K(\phi,X)-G(\phi,X)\Box\phi$5 (Nouri-Zonoz et al., 6 Nov 2025). A widely studied specialization is the shift-symmetric subclass $K(\phi,X)-G(\phi,X)\Box\phi$6, $K(\phi,X)-G(\phi,X)\Box\phi$7, for which the action is invariant under $K(\phi,X)-G(\phi,X)\Box\phi$8 and the scalar dynamics can be written in terms of a conserved Noether current (Bernardo et al., 2020).

The shift-symmetric current is model-dependent in detail because of sign conventions, but its covariant content is uniform: the scalar equation can be written as $K(\phi,X)-G(\phi,X)\Box\phi$9 when there is no explicit $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$0-dependence. In the conventions used in cosmological KGB models,

$$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$1

while in the stealth-black-hole convention it appears as

$$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$2

(Garcia-Arroyo et al., 2022, Bernardo et al., 2020). This conservation law is central both for attractor cosmology and for stealth sectors.

KGB is not restricted to shift symmetry. Static wormhole analyses, for example, treat the fully general $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$3, $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$4 theory, whereas post-GW170817 Jordan-frame constructions impose $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$5, $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$6, and allow a braiding term $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$7 together with $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$8 (Korolev et al., 2020, Racskó et al., 2020). This makes clear that “KGB” is sometimes used narrowly for minimally coupled $$S=\int d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R+K(\phi,X)-G(\phi,X)\,\Box\phi\right], \qquad X\equiv -\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi,$$9, and sometimes more broadly for cubic Horndeski sectors with braiding-type derivative structure.

A common misconception is that braiding necessarily implies anomalous tensor propagation. In the KGB sectors emphasized after GW170817, $G_4=M_{\rm Pl}^2/2$0 is constant or $G_4=M_{\rm Pl}^2/2$1-independent and $G_4=M_{\rm Pl}^2/2$2, so $G_4=M_{\rm Pl}^2/2$3 and tensor modes remain luminal (Garcia-Arroyo et al., 2022, Racskó et al., 2020).

2. Braiding, imperfect-fluid structure, and perturbative dynamics

A foundational feature of KGB is that the scalar stress tensor is not generically of perfect-fluid form. In the hydrodynamical formulation developed for KGB and related galileon sectors, one defines a vorticity-free four-velocity by

$G_4=M_{\rm Pl}^2/2$4

a chemical potential $G_4=M_{\rm Pl}^2/2$5, and a “diffusivity” $G_4=M_{\rm Pl}^2/2$6. The energy-momentum tensor then takes the imperfect-fluid form

$G_4=M_{\rm Pl}^2/2$7

with spatial energy flux

$G_4=M_{\rm Pl}^2/2$8

(Pujolas et al., 2011). The fluid is irrotational and non-dissipative, with exact EMT conservation and zero entropy production, but it remains imperfect because the energy flow is not aligned with either the scalar gradient or the full shift current (Pujolas et al., 2011).

On FRW backgrounds this structure simplifies, but braiding survives through explicit $G_4=M_{\rm Pl}^2/2$9-dependent terms in both the scalar current and the effective energy density. In the original dark-energy formulation of KGB, the Friedmann equation acquires a term linear in $G_5=0$0, and the scalar sector admits attractors in which the scalar “monitors” external matter through the current and the braiding function $G_5=0$1 (Deffayet et al., 2010). In shift-symmetric cosmology the current redshifts as $G_5=0$2, so $G_5=0$3 becomes a late-time attractor, but the location of that attractor depends on both the Lagrangian and the external energy density (Deffayet et al., 2010).

The perturbation theory is correspondingly modified. In the weak-field EFT parameterization used by relativistic $G_5=0$4-body work, only $G_5=0$5 and $G_5=0$6 are nonzero for KGB with constant Planck mass and luminal tensors, and the scalar variables

$G_5=0$7

enter the dark-energy density, pressure, and momentum perturbations through $G_5=0$8-dependent source terms (Nouri-Zonoz et al., 6 Nov 2025). Scalar viability is encoded in the no-ghost condition

$G_5=0$9

and in positivity of the scalar propagation speed

$(+---)$0

(Nouri-Zonoz et al., 6 Nov 2025). This formulation makes explicit that braiding is not merely a background effect; it directly changes the kinetic prefactor and the gravitational response of scalar perturbations.

3. Cosmological backgrounds, attractors, and reconstruction

KGB cosmology became prominent as a dark-energy framework because it can drive late-time acceleration, cross the phantom divide, and approach de Sitter without introducing extra DOF beyond one scalar and the massless spin-2 graviton (Deffayet et al., 2010). In one extensively studied shift-symmetric family,

$(+---)$1

the FRW attractor is obtained from $(+---)$2, which yields

$(+---)$3

This “nKGB” family interpolates between the cubic Galileon at $(+---)$4 and $(+---)$5CDM as $(+---)$6 in both background and linear perturbations (Garcia-Arroyo et al., 2022). Earlier large-scale-structure work on a related shift-symmetric model showed that the background expansion matches the Dvali–Turner form and reduces to $(+---)$7CDM for $(+---)$8, while the scalar sound speed and linear growth are correspondingly deformed at finite $(+---)$9 (Kimura et al., 2010).

KGB has also been used as an inflationary mechanism. In axion inflation with

$K(\phi,X)+G(\phi,X)\Box\phi$0

the braiding term provides additional friction, and the analysis of the paper shows a large parameter regime in which the axion decay constant can be naturally sub-Planckian while still producing an almost scale-invariant spectrum (Maity, 2012). This suggests that KGB can change the usual relation between axion field range and inflationary slow roll.

Beyond specifying $K(\phi,X)+G(\phi,X)\Box\phi$1 and $K(\phi,X)+G(\phi,X)\Box\phi$2 and solving forward, KGB also admits inverse construction. A reconstruction method for shift-symmetric flat FRW with nonzero conserved current chooses $K(\phi,X)+G(\phi,X)\Box\phi$3 and $K(\phi,X)+G(\phi,X)\Box\phi$4, then reconstructs $K(\phi,X)+G(\phi,X)\Box\phi$5 and $K(\phi,X)+G(\phi,X)\Box\phi$6. The method relies on

$K(\phi,X)+G(\phi,X)\Box\phi$7

and is illustrated for effective perfect-fluid backgrounds, a unified dark-energy–dark-matter expansion history, and a post-inflationary transition to radiation domination (Muharlyamov et al., 2021). This indicates that KGB is not only a model class but also a reconstruction framework within Horndeski.

4. Large-scale structure, neutrino degeneracies, and numerical modeling

At the level of linear observables, KGB modifies the growth of structure through a braiding-induced fifth force while often leaving intrinsic gravitational slip negligible when $K(\phi,X)+G(\phi,X)\Box\phi$8 is constant and $K(\phi,X)+G(\phi,X)\Box\phi$9 (Garcia-Arroyo et al., 2022). In the nKGB models analyzed with hi_class and MontePython, the fifth force enhances power while massive neutrinos suppress it, producing a partial degeneracy at intermediate linear scales. However, the authors find a distinctive large-scale bump in the matter power spectrum around $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$0–$X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$1 that survives neutrino-mass tuning: at $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$2 the bump is approximately $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$3 for $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$4 and approximately $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$5 for $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$6, while at $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$7 a residual $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$8 bump remains for $X=\frac12 g^{\mu\nu}\nabla_\mu\phi\nabla_\nu\phi$9 even after tuning $G\Box\phi$0 (Garcia-Arroyo et al., 2022). The same study reports that $G\Box\phi$1 shows no $G\Box\phi$2 tension, all studied nKGB cases show no $G\Box\phi$3 tension, and a null neutrino mass is excluded in that framework (Garcia-Arroyo et al., 2022).

An earlier observational discriminator was the ISW–LSS cross-correlation. For the specific shift-symmetric model with $G\Box\phi$4 and $G\Box\phi$5, the late-time ISW effect anti-correlates with large-scale structure over a wide parameter range because the enhanced effective gravitational coupling makes the metric potentials grow rather than decay. Using the six-catalogue compilation of Giannantonio et al., the paper derives the bound

$G\Box\phi$6

thereby ruling out the covariant Galileon case $G\Box\phi$7 in that model (Kimura et al., 2011). A common misconception is therefore that any shift-symmetric KGB background close to $G\Box\phi$8CDM automatically remains viable at the perturbative level; ISW data showed that the sign of potential evolution can be decisive.

Fully relativistic simulation of KGB has now become possible. “KGB-evolution” extends gevolution by solving the linearized KGB dark-energy equations in a nonlinear metric and matter background, with $G\Box\phi$9 amplifying dark-energy clustering relative to the $K$0-essence limit (Nouri-Zonoz et al., 6 Nov 2025). The code validates against hi_class with sub-percent agreement in the appropriate linear regimes, and in a stress-test model with $K$1, $K$2, $K$3, $K$4, nonlinear evolution amplifies the matter-spectrum difference between KGB and $K$5-essence to approximately $K$6 at $K$7 (Nouri-Zonoz et al., 6 Nov 2025). This suggests that braiding may be more strongly constrained by lensing and ISW observables than by matter power alone, because the potentials respond directly to $K$8 and $K$9.

5. Strong-gravity sectors: stealth black holes, wormholes, shells, and lower-dimensional realizations

KGB has a rich strong-gravity sector. In shift-symmetric, luminal KGB, all hairy stealth black holes arise when the scalar kinetic density is covariantly constant, $K(\phi,X)-G(\phi,X)\Box\phi$00, and the theory satisfies

$K(\phi,X)-G(\phi,X)\Box\phi$01

so that the metric is exactly Kerr–(A)dS or its static limits while the scalar hair remains nontrivial (Bernardo et al., 2019). This construction exploits a loophole in a previous no-go claim: shift symmetry need not be broken if $K(\phi,X)-G(\phi,X)\Box\phi$02 is covariantly constant (Bernardo et al., 2019).

Linear perturbations about static stealth black holes sharpen the interpretation of braiding in this regime. For asymptotically flat or Schwarzschild–(A)dS stealth solutions, odd-parity perturbations satisfy the Regge–Wheeler equation with the same potential as GR, whereas the even-parity Zerilli equation acquires an additional source term from the scalar hair,

$K(\phi,X)-G(\phi,X)\Box\phi$03

or $K(\phi,X)-G(\phi,X)\Box\phi$04 times the same radial factor in the explicit $K(\phi,X)-G(\phi,X)\Box\phi$05-essence example (Bernardo et al., 2020). The monopole and dipole sectors admit exact solutions, but for $K(\phi,X)-G(\phi,X)\Box\phi$06 they are typically pathological non-gauge modes, diverging at horizons or at large radius depending on the asymptotics; when $K(\phi,X)-G(\phi,X)\Box\phi$07, the low multipoles reduce to the pure-gauge GR shifts (Bernardo et al., 2020). This pathology is traced to the scalar mode becoming non-propagating, with an effective sound speed formally tending to infinity on stealth backgrounds (Bernardo et al., 2020).

The same derivative structure that enables stealth sectors also allows traversable wormholes. In static, spherically symmetric KGB wormholes with metric

$K(\phi,X)-G(\phi,X)\Box\phi$08

the throat conditions imply, among other constraints,

$K(\phi,X)-G(\phi,X)\Box\phi$09

and $K(\phi,X)-G(\phi,X)\Box\phi$10 requires $K(\phi,X)-G(\phi,X)\Box\phi$11 (Korolev et al., 2020). The paper constructs an analytic Ellis–Bronnikov-like wormhole for $K(\phi,X)-G(\phi,X)\Box\phi$12, $K(\phi,X)-G(\phi,X)\Box\phi$13, and numerical asymptotically AdS wormholes for $K(\phi,X)-G(\phi,X)\Box\phi$14, $K(\phi,X)-G(\phi,X)\Box\phi$15, with a near-throat “gravitational barrier” controlled by the conserved braiding current $K(\phi,X)-G(\phi,X)\Box\phi$16 (Korolev et al., 2020).

KGB also alters hypersurface matching. In a GW-safe cosmologically viable Jordan-frame subclass with

$K(\phi,X)-G(\phi,X)\Box\phi$17

the Barrabès–Israel null-shell relations are generalized to

$K(\phi,X)-G(\phi,X)\Box\phi$18

$K(\phi,X)-G(\phi,X)\Box\phi$19

$K(\phi,X)-G(\phi,X)\Box\phi$20

together with an additional scalar junction condition absent in GR (Racskó et al., 2020).

In two dimensions the relation between KGB and generalized dilaton gravity becomes exact: the most general 2D scalar–tensor theory with second-order EL equations reduces to KGB, and the nonminimal coupling $K(\phi,X)-G(\phi,X)\Box\phi$21 can be rewritten through $K(\phi,X)-G(\phi,X)\Box\phi$22 and $K(\phi,X)-G(\phi,X)\Box\phi$23 involving $K(\phi,X)-G(\phi,X)\Box\phi$24 (Takahashi et al., 2018). Shift-symmetric 2D KGB even admits a complete classification of static solutions with $K(\phi,X)-G(\phi,X)\Box\phi$25, where $K(\phi,X)-G(\phi,X)\Box\phi$26 is constant and the metric function is quadratic in the spatial coordinate (Takahashi et al., 2018).

6. Self-tuning, singular futures, and recurring controversies

One major line of work concerns whether shift symmetry in KGB forces a de Sitter future. In the self-tuning, tadpole-free, shift-symmetric sector,

$K(\phi,X)-G(\phi,X)\Box\phi$27

the paper on self-tuning KGB shows that well-tempering is impossible without a tadpole, but a trivial-scalar self-tuning branch exists and drives the system to a stable de Sitter state $K(\phi,X)-G(\phi,X)\Box\phi$28. On the self-tuned vacuum, ghost and gradient stability reduce to the band

$K(\phi,X)-G(\phi,X)\Box\phi$29

and the shift current hypersurface $K(\phi,X)-G(\phi,X)\Box\phi$30 coincides with the de Sitter attractor in the healthy branch (Bernardo, 2021). This supports the view that KGB can self-adjust to acceleration while preserving $K(\phi,X)-G(\phi,X)\Box\phi$31.

That conclusion is not generic. In the pure-braiding shift-symmetric model

$K(\phi,X)-G(\phi,X)\Box\phi$32

a big-rip future arises for

$K(\phi,X)-G(\phi,X)\Box\phi$33

because the late-time Hubble rate scales as a positive power of $K(\phi,X)-G(\phi,X)\Box\phi$34, so both $K(\phi,X)-G(\phi,X)\Box\phi$35 and $K(\phi,X)-G(\phi,X)\Box\phi$36 diverge in finite cosmic time (Vasilev et al., 2022). The paper gives an illustrative evolution with Planck-like parameters and $K(\phi,X)-G(\phi,X)\Box\phi$37, for which matter–scalar equality occurs at $K(\phi,X)-G(\phi,X)\Box\phi$38, the scalar becomes phantom at $K(\phi,X)-G(\phi,X)\Box\phi$39, $K(\phi,X)-G(\phi,X)\Box\phi$40 falls below $K(\phi,X)-G(\phi,X)\Box\phi$41 only at $K(\phi,X)-G(\phi,X)\Box\phi$42, and the big rip occurs in approximately $K(\phi,X)-G(\phi,X)\Box\phi$43 Gyr (Vasilev et al., 2022). A broader dynamical-systems analysis extends this picture: power-law $K(\phi,X)-G(\phi,X)\Box\phi$44 and $K(\phi,X)-G(\phi,X)\Box\phi$45 sectors admit not only self-tuning de Sitter points but also big-rip, big-freeze, and sudden-singularity attractors depending on the exponents (Vasilev et al., 2022).

A second controversy concerns perturbative health in extended measure frameworks. In two-field measure theory, k-essence remains potentially viable, but KGB becomes far more pathological: the scalar perturbations are luminal, $K(\phi,X)-G(\phi,X)\Box\phi$46, yet one mode is always a ghost, and explicit examples display tachyonic instabilities associated with the canonical sector (Cordero et al., 2022). This suggests that some apparently benign extensions of KGB radically alter the scalar kinetic matrix.

Two recurring misconceptions therefore deserve emphasis. First, KGB is not synonymous with stable phantom acceleration: the theory can cross $K(\phi,X)-G(\phi,X)\Box\phi$47 without ghosts in suitable models, but explicit shift-symmetric sectors also admit unstable phantom attractors and finite-time singularities (Deffayet et al., 2010, Vasilev et al., 2022). Second, strong coupling is not an incidental detail confined to exotic solutions: stealth black holes, self-tuning sectors, and measure-theory variants all show that scalar propagation can degenerate or become nonhyperbolic in precisely the regimes where braiding most strongly alters the background (Bernardo et al., 2020, Cordero et al., 2022).

KGB thus occupies a distinctive position within Horndeski theory. It is the minimal derivative scalar–tensor sector that preserves second-order dynamics while generating genuine kinetic mixing, and it provides a unified arena in which imperfect-fluid dark energy, modified growth, screening, stealth geometry, nontrivial junction conditions, and self-tuning can all be studied within the same pair of functions $K(\phi,X)-G(\phi,X)\Box\phi$48 and $K(\phi,X)-G(\phi,X)\Box\phi$49. The same economy of structure that makes KGB analytically tractable also makes its pathologies unusually transparent.