Kinnersley Tetrads and Newman–Penrose Curvature

Kerr perturbations become tractable when curvature is resolved along the repeated principal null directions of the type-D background. The Kinnersley tetrad makes that alignment explicit: the unperturbed extreme Weyl scalars vanish, so their first-order perturbations isolate radiative curvature without first solving for ten metric components. The price is convention sensitivity. Null-vector normalization, Weyl-scalar signs, spin-boost weights, and horizon regularity must all be fixed before a separated component can be given physical meaning.

Required background. Frames, Coframes, and Anholonomy supplies tetrad gauge freedom, null frames, and frame-component calculations.

Helpful background. Black-Hole Geometry and Mechanics supplies the Kerr exterior and its horizon generator; PDEs and Green Functions for Gravitation supplies characteristic data and boundary conditions; Formulations of General Relativity supplies gauge, constraints, and characteristic formulations.

Principal null tetrad on Kerr

Work in the subextremal Kerr exterior, with $M\gt0$, $\lvert a\rvert\lt GM$, and $r\gt r_+=GM+\sqrt{(GM)^2-a^2}$, in Boyer–Lindquist coordinates $(t,r,\theta,\phi)$. Here $M$ is the physical mass, $a=J/M$ has dimensions of length, and

$$\Sigma=r^2+a^2\cos^2\theta, \qquad \Delta=r^2-2GMr+a^2.$$

The exact geometry and its maximal extension belong to the Kerr geometry pages; here it is a fixed vacuum background. With the site signature $(-,+,+,+)$, order the null tetrad as $(\ell^a,n^a,m^a,\bar m^a)$ and normalize

$$\ell\!\cdot n=-1, \qquad m\!\cdot\bar m=1,$$

with every other inner product zero. A Kinnersley choice is

$$\begin{aligned} \ell^\mu&=\left(\frac{r^2+a^2}{\Delta},1,0,\frac{a}{\Delta}\right),\\ n^\mu&=\frac{1}{2\Sigma}\left(r^2+a^2,-\Delta,0,a\right),\\ m^\mu&=\frac{1}{\sqrt2\,(r+ia\cos\theta)} \left(ia\sin\theta,0,1,\frac{i}{\sin\theta}\right). \end{aligned}$$

Direct contraction with the Kerr metric gives the stated inner products and reconstructs

$$g^{ab}=-2\ell^{(a}n^{b)}+2m^{(a}\bar m^{b)}.$$

Teukolsky’s original paper uses the opposite signature, for which the same displayed vector components obey $\ell\!\cdot n=1$ and $m\!\cdot\bar m=-1$ Teukolsky 1973, § IV, Eqs. (4.2)–(4.4), printed p. 640. Importing its Newman–Penrose equations therefore requires translating the metric, curvature definitions, and all contractions together. In particular, an overall metric-sign change leaves $R^a{}_{bcd}$ fixed but reverses the all-lowered Weyl tensor; changing one sign in isolation is not a valid conversion.

Spin coefficients and type-D curvature

Let $D=\ell^a\nabla_a$, $\mathord\Delta_{\!NP}=n^a\nabla_a$, and $\delta=m^a\nabla_a$, with $\bar\delta=\bar m^a\nabla_a$. The label on $\mathord\Delta_{\!NP}$ distinguishes the directional derivative from the Kerr polynomial $\Delta(r)$. A bar on a coefficient below denotes complex conjugation.

The signs of spin-coefficient symbols alone are not universal across NP references. To make the convention reproducible, define the connection contraction

$$\Gamma(X,Y,Z)=X^aY^b\nabla_b Z_a,$$

where $Z_a=g_{ab}Z^b$ is lowered with the site’s metric. Use

$$\begin{aligned} \kappa&=-\Gamma(m,\ell,\ell),& \sigma&=-\Gamma(m,m,\ell),\\ \rho_{NP}&=-\Gamma(m,\bar m,\ell),& \tau&=-\Gamma(m,n,\ell),\\ \pi&=\Gamma(\bar m,\ell,n),& \mu_{NP}&=\Gamma(\bar m,m,n),\\ \lambda_{NP}&=\Gamma(\bar m,\bar m,n),& \nu&=\Gamma(\bar m,n,n). \end{aligned}$$

The four remaining coefficients are

$$\begin{aligned} \epsilon&=\frac12[\Gamma(\bar m,\ell,m)-\Gamma(n,\ell,\ell)],\\ \gamma&=\frac12[\Gamma(\bar m,n,m)-\Gamma(n,n,\ell)],\\ \beta&=\frac12[\Gamma(\bar m,m,m)-\Gamma(n,m,\ell)],\\ \alpha&=\frac12[\Gamma(\bar m,\bar m,m)-\Gamma(n,\bar m,\ell)]. \end{aligned}$$

Direct differentiation of the displayed tetrad gives

$$\kappa=\sigma=\lambda_{NP}=\nu=\epsilon=0.$$

Thus both real null congruences are geodesic and shear-free. Writing $\rho=\rho_{NP}$ temporarily, the nonzero coefficients are

$$\begin{aligned} \rho&=-\frac{1}{r-ia\cos\theta},& \beta&=-\frac{\bar\rho\cot\theta}{2\sqrt2},\\ \pi&=\frac{ia\rho^2\sin\theta}{\sqrt2},& \tau&=-\frac{ia\rho\bar\rho\sin\theta}{\sqrt2},\\ \mu_{NP}&=\frac{\rho^2\bar\rho\Delta}{2},& \gamma&=\mu_{NP}+\frac{\rho\bar\rho(r-GM)}2,\\ \alpha&=\pi-\bar\beta. \end{aligned}$$

The plus sign between the two terms in $\gamma$ is consequential. These contraction definitions retain the numerical coefficient values in Teukolsky 1973, p. 640, Eq. (4.5) after translating the signature. For example, at $a=0$ they give $\rho=-1/r$, $\mu_{NP}=-f/(2r)$ and $\gamma=GM/(2r^2)$, where $f=1-2GM/r$. Every coefficient has inverse-length dimension.

With the Weyl definitions

$$\Psi_0=-C_{abcd}\ell^a m^b\ell^c m^d, \qquad \Psi_1=-C_{abcd}\ell^a n^b\ell^c m^d,$$ $$\Psi_2=-C_{abcd}\ell^a m^b\bar m^c n^d, \qquad \Psi_4=-C_{abcd}n^a\bar m^b n^c\bar m^d,$$

and the analogous $\Psi_3$, the Kerr background has

$$\Psi_0=\Psi_1=\Psi_3=\Psi_4=0, \qquad \Psi_2=+\frac{GM}{(r-ia\cos\theta)^3}.$$

This is the operational content of Petrov type D for the calculation. Teukolsky writes $\rho_{NP}=-1/(r-ia\cos\theta)$ and $\Psi_2^{\rm T}=GM\rho_{NP}^3$, so his source-local scalar is negative; the complete signature and curvature translation above gives $\Psi_2=-\Psi_2^{\rm T}$ in the site’s package Teukolsky 1973, § IV, Eqs. (4.5)–(4.6), printed p. 640. Substitution into the Newman–Penrose Ricci and Bianchi identities is an independent check of the tetrad, curvature sign, and spin coefficients. Teukolsky uses precisely these background zeros to eliminate coupled perturbed quantities and obtain equations for the two extreme components Teukolsky 1973, § II, Eqs. (2.2)–(2.15), printed pp. 637–638.

Two useful identities can already be checked directly:

$$D\Psi_2=3\rho_{NP}\Psi_2,\qquad \delta\Psi_2=3\tau\Psi_2.$$

The background scalar depends only on $r,\theta$, so the first identity reduces to a radial derivative; the second uses the angular component of $m^a$. They supply the product-rule terms in the explicit vacuum elimination. The signature translation must also be applied to the perturbed Ricci identity there; changing only the final background Weyl symbol loses its relative sign.

Curvature and tetrad freedom

Write the metric and Weyl tensor as $g_{ab}+\varepsilon h_{ab}$ and $C_{abcd}+\varepsilon\,\delta C_{abcd}+O(\varepsilon^2)$. Because the background extreme scalars vanish, $\delta\Psi_0$ and $\delta\Psi_4$ are invariant under infinitesimal coordinate gauge transformations. They are also invariant at first order under infinitesimal tetrad rotations about the fixed background tetrad in the sense proved in the original construction Teukolsky 1973, Appendix A, pp. 646–647. This does not make them independent of a finite spin-boost convention.

The residual transformation preserving the principal directions is

$$\ell^a\mapsto A\ell^a, \quad n^a\mapsto A^{-1}n^a, \quad m^a\mapsto e^{i\chi}m^a,$$

where $A>0$ and $\chi$ are real. Consequently,

$$\delta\Psi_0\mapsto A^2e^{2i\chi}\delta\Psi_0, \qquad \delta\Psi_4\mapsto A^{-2}e^{-2i\chi}\delta\Psi_4.$$

They are curvature quantities with definite boost and spin weights, not ordinary tetrad-independent scalars. The two components also emphasize different null directions: $\delta\Psi_4$ is normally adapted to outgoing radiation at future null infinity, whereas $\delta\Psi_0$ is adapted to the complementary characteristic direction. Flux or strain requires asymptotic normalization and reconstruction; it is not equal to a raw Weyl component.

Horizon-regular interpretation

The displayed Kinnersley tetrad is singular at $\Delta=0$ in Boyer–Lindquist components. That is a chart-and-frame singularity, not a curvature conclusion. To test future-horizon regularity, transform to ingoing Kerr coordinates and map the displayed Kinnersley frame to a regular frame with a spin-boost $A=C\Delta$, where $C$ is smooth and nonzero at the horizon. A standard choice is $C=[2(r^2+a^2)]^{-1}$, up to further smooth finite rescalings Teukolsky and Press 1974, §§II and IV, pp. 444–451. In this direction of the map,

$$\delta\Psi_0^{\mathrm{reg}} =C^2\Delta^2e^{2i\chi}\delta\Psi_0^{\mathrm{Kin}}, \qquad \delta\Psi_4^{\mathrm{reg}} =C^{-2}\Delta^{-2}e^{-2i\chi}\delta\Psi_4^{\mathrm{Kin}}.$$

Thus a regular $\delta\Psi_4^{\mathrm{reg}}$ requires $\delta\Psi_4^{\mathrm{Kin}}=O(\Delta^2)$ in this convention. The factor $\Delta^{-2}$ belongs to the frame conversion; it is not by itself a curvature divergence.

The physical horizon phase is fixed by the future generator

$$\chi_H=\partial_t+\Omega_H\partial_\phi, \qquad \Omega_H=\frac{a}{r_+^2+a^2}.$$

For the chapter convention $e^{-i\omega t+im\phi}$, a mode has generator frequency $k_H=\omega-m\Omega_H$. A future-horizon ingoing field is smooth in an advanced, co-rotating chart; the sign of a Boyer–Lindquist exponential alone is insufficient without the tetrad weight. Teukolsky and Press explicitly compare Kinnersley and horizon-adapted tetrads and show why distinct field representations are useful in distinct asymptotic regions Teukolsky–Press 1974, §§II and IV, pp. 444–451.

In the Schwarzschild limit $a\to0$, the tetrad becomes a principal radial null tetrad, $\Psi_2\to+GM/r^3$, and the two radiative components reduce to the familiar spin-weight $\pm2$ curvature variables. This limit checks the mass dimension, phase-free background curvature, and interchange symmetry of the principal directions.

Common pitfalls

Calling a Kinnersley component horizon singular. Boyer–Lindquist coordinates and the unboosted principal tetrad are both singular at the future horizon. Transform coordinates and tetrad, then test reconstructed curvature or regular-frame components.

Calling $\delta\Psi_4$ the waveform. Its radiative interpretation depends on tetrad normalization and asymptotic limit. At large radius it is related to two time derivatives of strain only after fixing orientation, polarization dyad, retarded time, and normalization.

Discarding the nonradiative sector. Vanishing $\delta\Psi_0$ and $\delta\Psi_4$ does not force $h_{ab}$ to be pure gauge. Variations of Kerr mass and angular momentum are the essential completion counterexamples developed later in this chapter.

Exercises

Verify three independent checks on the displayed tetrad: its spin-boost weights, its Schwarzschild curvature limit, and the interpretation of an apparent $\Delta^{-2}$ divergence in $\delta\Psi_4$ after a boost $A\propto\Delta$.

Solution

Each occurrence of $\ell$ contributes $A e^{0i\chi}$ and each occurrence of $m$ contributes $e^{i\chi}$, so $\delta\Psi_0$, containing two of each, transforms as $A^2e^{2i\chi}\delta\Psi_0$. Likewise two $n$ and two $\bar m$ give $A^{-2}e^{-2i\chi}\delta\Psi_4$. The normalization remains fixed because $(A\ell)\!\cdot(A^{-1}n)=-1$ and $(e^{i\chi}m)\!\cdot(e^{-i\chi}\bar m)=1$.

Setting $a=0$ gives $\Sigma=r^2$, $\Delta=r(r-2GM)$, $m^a=(0,0,1,i/\sin\theta)/(\sqrt2r)$, and $\Psi_2=+GM/r^3$. These are the Schwarzschild principal-null and tidal-curvature values in the declared site convention.

Finally, take the displayed Kinnersley frame as the old frame and the ingoing regular frame as the new one. With $A=C\Delta$ and smooth $\chi$, $\delta\Psi_4^{\mathrm{reg}}=C^{-2}\Delta^{-2}e^{-2i\chi} \delta\Psi_4^{\mathrm{Kin}}$. Therefore a finite nonzero regular-frame limit corresponds to $\delta\Psi_4^{\mathrm{Kin}}=O(\Delta^2)$; a Kinnersley component approaching a nonzero constant would instead produce a genuine regular-frame divergence. Reversing the frame map inverts these factors. The invariant diagnosis is to write the actual regular tetrad in ingoing coordinates and contract the Weyl tensor with it, not to inspect an isolated power of $\Delta$.

References

• Teukolsky, Saul A. “Perturbations of a Rotating Black Hole. I. Fundamental Equations for Gravitational, Electromagnetic, and Neutrino-Field Perturbations.” Astrophysical Journal 185 (1973): 635–647. DOI.
• Teukolsky, Saul A., and William H. Press. “Perturbations of a Rotating Black Hole. III. Interaction of the Hole with Gravitational and Electromagnetic Radiation.” Astrophysical Journal 193 (1974): 443–461. ADS record and PDF.