Kinetic and Derivative Screening

Kinetic screening suppresses a scalar force when a large first-derivative invariant makes fluctuations stiff. It differs from chameleon screening, which changes an effective potential, and from cubic Vainshtein screening, which is governed by second derivatives. The mechanism can reduce the force even while the scalar gradient is large, but only on branches whose effective characteristic metric remains Lorentzian.

Required background. Ordinary Differential Equations supplies radial first integrals and asymptotic balances.

Helpful background. Einstein Equations supplies the GR force used for comparison; Gravitational Actions and Boundary Terms supplies the nonlinear field variation; Formulations of General Relativity supplies characteristics and hyperbolicity; Gravity Beyond General Relativity supplies EFT and branch checks; and Relativistic Matter and Compact Objects supplies sources for which pressure, compactness, and time dependence matter.

A kinetic scalar coupled to matter

Use $M_{\rm Pl}^{-2}=8\pi G$, $\Lambda\gt0$, and define

$$\chi=-\frac{\nabla_\mu\phi\nabla^\mu\phi}{2\Lambda^4}$$

and take

$$S=\int d^4x\sqrt{-g} \left[ \frac{M_{\rm Pl}^2}{2}R+\Lambda^4K(\chi) \right] +S_m[A^2(\phi)g_{\mu\nu},\psi],$$

with $A(\phi)\simeq1+\beta\phi/M_{\rm Pl}$. The scalar equation is

$$\nabla_\mu\left(K_\chi\nabla^\mu\phi\right) =-\frac{\beta T}{M_{\rm Pl}}.$$

For ordinary nonrelativistic matter, $T\simeq-\rho$, so the source on the right is positive for $\beta\gt0$. The weak-field canonical limit requires $K_\chi\to1$ as $\chi\to0$. A static spatial gradient has $\chi\lt0$ with the site signature; a homogeneous timelike gradient has $\chi\gt0$. A function healthy on one side of $\chi=0$ need not be healthy on the other.

Spherical screening from a large kinetic coefficient

For a static spherical source, integrate once outside the matter:

$$r^2K_\chi(\chi)\phi'=C, \qquad C=\frac{\beta M}{4\pi M_{\rm Pl}}, \qquad \chi=-\frac{(\phi')^2}{2\Lambda^4}.$$

The scalar-to-Newton force ratio is

$$\frac{F_\phi}{F_N} =\frac{2\beta^2}{K_\chi}.$$

This equation displays the mechanism directly. Far from the source, $K_\chi\simeq1$ and the scalar is unscreened. Where the branch reaches $K_\chi\gg1$, the same conserved radial flux is carried by a smaller gradient than its linear extrapolation, and the force ratio decreases.

For a power-law large-gradient branch,

$$K_\chi\sim(-\chi)^{m-1}, \qquad m\gt1,$$

the first integral gives

$$\phi'\propto r^{-2/(2m-1)}.$$

Because $2/(2m-1)\lt2$, this gradient grows inward more slowly than the linear $r^{-2}$ solution. The kinetic radius is set parametrically by the linear solution reaching $|\chi|\sim1$:

$$r_K^2\sim\frac{C}{\Lambda^2}.$$

Order-one factors depend on the normalization of $K$. The construction of such derivative-screened branches is introduced in Babichev, Deffayet, and Ziour 2009, equations (13)–(20).

Fluctuation metric and characteristic speeds

Set $\phi=\bar\phi+\varphi$. The principal quadratic operator is

$$\mathcal Z^{\mu\nu}\nabla_\mu\nabla_\nu\varphi, \qquad \mathcal Z^{\mu\nu} =K_\chi g^{\mu\nu} -\frac{K_{\chi\chi}}{\Lambda^4} \nabla^\mu\bar\phi\nabla^\nu\bar\phi.$$

On a static radial background, the tangential kinetic eigenvalue is $K_\chi$, while the radial one is

$$K_\chi+2\chi K_{\chi\chi}.$$

Both must have the appropriate positive sign for a hyperbolic, ghost-free branch. Their ratio controls the radial characteristic speed. Screening $K_\chi\gg1$ does not ensure that the second combination is positive, finite, or subluminal. A point at which an eigenvalue vanishes is a characteristic degeneracy, not a smooth recovery of GR.

This effective-metric test is stronger than checking the background profile: it asks whether small disturbances have a well-posed initial-value problem in the same environment used to claim screening.

For the power-law example above,

$$K_\chi+2\chi K_{\chi\chi}=(2m-1)K_\chi, \qquad c_r^2=2m-1,\qquad c_\perp^2=1.$$

Thus $m\gt1$ screens the static force with healthy principal signs, but its radial cone is wider than the matter cone. On a homogeneous timelike background the same two combinations exchange temporal and longitudinal roles, giving $c_s^2=K_\chi/(K_\chi+2\chi K_{\chi\chi})$ when both are positive. One must evaluate the function on the appropriate sign of $\chi$. A static screening calculation does not determine the cosmological sound speed.

Caustics, time dependence, and matching

The spherical algebraic relation can possess several roots. The physical branch must connect continuously to $K_\chi\to1$ in the far zone and remain regular through the transition. A cosmological timelike background shifts $\chi$ before the local spatial gradient is added; multiple sources destroy exact spherical reduction; and rapid source evolution can invalidate the quasistatic first integral.

Nonlinear kinetic waves may steepen and form caustics in some $P(X)$ theories. Whether that occurs depends on $K$, data class, and symmetry; it is not decided by the static solution. The model must also keep higher-derivative operators below the background-dependent cutoff. Joyce et al. compare first-derivative and second-derivative screening and their EFT regimes in Joyce et al. 2015, §§ 6.1–6.2.

Mechanism comparison

Mechanism	Nonlinear control variable	What suppresses the exterior response	Principal failure check
Chameleon or symmetron	Field value and effective potential	Large mass, thin charge shell, or vanishing coupling	Boundary-dependent minimum and quantum control
Kinetic screening	First derivative $(\nabla\phi)^2$	Large $K_\chi$ reduces the force and fluctuation coupling	Effective-metric signature, caustics, and branch matching
Vainshtein screening	Second derivatives $\nabla\nabla\phi$	Large Hessian-dependent kinetic matrix	Characteristic cone, strong coupling, and nonspherical matching

Calling all three “nonlinear screening” does not permit their screening radii, charges, or equivalence-principle predictions to be exchanged.

An independent GR-limit check sends $\beta\to0$ at fixed regular background: the matter source then drops out of the scalar equation and $F_\phi/F_N\to0$. Sending $K_\chi\to\infty$ is not the same limit; it retains a nontrivial background whose stress energy and characteristic cone still have to be included in the metric equations.

Exercise

Assume $K_\chi=(-\chi)^{m-1}$ on a static screened branch with $m\gt1$. Derive the radial power of $\phi'$ and show how the fifth-force ratio changes inward. State one condition the background solution does not test.

Solution

Because $-\chi=(\phi')^2/(2\Lambda^4)$, the first integral gives $r^2(\phi')^{2m-2}\phi'=\text{constant}$. Therefore $\phi'\propto r^{-2/(2m-1)}$. The linear gradient scales as $r^{-2}$, so their ratio and hence $F_\phi/F_N$ scale as $r^{\,2-2/(2m-1)}$, decreasing toward smaller $r$. This confirms screening for the selected root. It does not check hyperbolicity: one must separately evaluate $K_\chi$ and $K_\chi+2\chi K_{\chi\chi}$ and verify that the effective metric has the correct Lorentzian signature throughout the profile.

For this particular power law the check can be completed: $K_\chi+2\chi K_{\chi\chi}=(2m-1)K_\chi\gt0$, so $c_r^2=2m-1\gt1$ in the static large-gradient regime. The force is screened even though radial scalar disturbances propagate faster than the matter light speed in this effective description.

References

• Babichev, Eugeny, Cédric Deffayet, and Riad Ziour (2009), “k-Mouflage Gravity,” arXiv:0905.2943. Open PDF.
• Joyce, Austin, Bhuvnesh Jain, Justin Khoury, and Mark Trodden (2015), “Beyond the Cosmological Standard Model,” Physics Reports 568, 1–98. DOI. Open PDF.