Kerr Black Hole Ringdown in Effective Field Theory William L. Boyce, Jorge E. Santos Abstract We develop a systematic effective field theory calculation of the quasinormal modes of Kerr black holes valid for arbitrary spin, providing model-independent corrections to their ringdown spectrum directly relevant for gravitational-wave observations. Close to extremality, the effective field theory corrections in the grand‑canonical ensemble exhibit an oscillatory dependence on log ⁡ τ H \log\tau_{H} , with τ H ≡ T H / Ω H \tau_{H}\equiv T_{H}/\Omega_{H} a dimensionless measure of the black hole temperature. This behaviour signals an underlying discrete‑scale‑invariant structure. Introduction. General Relativity (GR) continues to exhibit remarkable success in accounting for the rapidly expanding body of observational data produced by the LIGO–Virgo–KAGRA (LVK) Collaboration during its recent observing runs 48 . At the same time, black‑hole physics has consistently driven progress in our understanding of gravity, and the increasing precision of compact‑object observations promises to sharpen this role even further. Yet even with these successes, GR cannot be the final word: it predicts physical singularities 56 , offers no mechanism to reconcile gravity with quantum field theory 43 , and leaves unresolved both the origin of black‑hole entropy 8 ; 64 and the fate of information in black‑hole evaporation 39 . These long‑standing issues point inexorably toward the need for a quantum theory of gravity. In view of these open problems, we adopt an agnostic stance toward the ultraviolet completion of gravity and work within the framework of effective field theory (EFT). EFT provides the most general and systematically improvable description of low‑energy deviations from GR, encoding all allowed corrections as higher‑dimensional operators suppressed by a heavy scale and constrained by locality, unitarity, Lorentz invariance, analyticity, and, crucially for gravity, diffeomorphism invariance. These principles ensure that any extension of Einstein gravity remains predictive 35 ; 36 while keeping assumptions about the underlying UV theory to a minimum. At the same time, EFT has well‑defined limits: it cannot capture intrinsically non‑local proposals—such as island constructions 3 ; 1 ; 4 ; 55 ; 2 ; 5 , fuzzball microstate structure 49 ; 9 ; 10 ; 50 ; 11 ; 40 ; 59 ; 41 , or other horizon‑scale non‑local effects—which lie outside the domain of validity of a local derivative expansion. Although EFT coefficients are often assumed to be minuscule—suppressed by appropriate powers of the Planck scale—their actual size depends sensitively on the spectrum of heavy fields that has been integrated out. As a result, these coefficients encode valuable information about possible ultraviolet matter sectors and can place meaningful constraints on hypothetical new physics. Moreover, their impact is not uniformly small: near extremality, parity‑even higher‑derivative operators can generate parametrically enhanced corrections, producing sizeable deviations from General‑Relativistic predictions while remaining fully consistent with EFT power counting 44 ; 45 . In this work, we examine an important component of this broader effort: the quasinormal‑mode (QNM) spectrum of rotating black holes, of the type routinely observed by the LVK Collaboration. The EFT description of Schwarzschild black holes—including their QNMs—was pioneered in 23 , whereas the EFT description of the QNM spectra of Reissner–Nordström black holes was developed only recently in 12 . However, much of the existing literature incorporates beyond‑GR corrections using a small‑spin expansion 22 ; 19 ; 21 ; 20 ; 14 ; 6 1 1 1 A notable exception is 52 , which studies the QNMs of electromagnetic waves in an EFT framework on a fixed Kerr background. . We demonstrate that this expansion fails catastrophically once the dimensionless spin exceeds a critical value, corresponding to the point at which the specific heat of a Kerr black hole changes sign. Although theoretical considerations already suggest why such a breakdown might occur, it is the direct numerical comparison with the small‑spin expansion that makes the failure unmistakable. These results highlight that accurate modelling of high‑spin black‑hole spectroscopy—precisely the regime probed by many LVK detections—requires techniques that extend beyond the traditional low‑spin approximation. We further note that approaches relying solely on the eikonal limit introduce additional subtleties 17 ; 15 ; 18 ; 16 . While the eikonal approximation captures certain high‑angular‑momentum features, subleading corrections—including those induced by higher‑derivative interactions—can contribute at the same order as the leading eikonal term, undermining the reliability of the expansion (see Sec. 4.8 of 58 for a detailed discussion). Consequently, the eikonal regime can become overly sensitive to the structure of the underlying EFT, and careless extrapolation may lead to qualitatively incorrect conclusions about the QNM spectrum. Understanding these limitations is therefore essential when using QNMs to test or constrain extensions of GR. Finally, we emphasise that the methods developed in this work rely solely on the EFT framework and require no assumptions about the ultraviolet completion of gravity. Our analysis remains valid across the entire range of black‑hole spins, including the high‑spin regime where traditional approximations fail. This ensures that our results provide a fully controlled and model‑independent characterisation of quasinormal‑mode physics within the domain of applicability of EFT. The EFTs. The most general EFT extension of the Einstein-Hilbert action involving up to eight derivatives is 33 S=\frac{1}{2\kappa_{\rm 4D}^{2}}\int\sqrt{-g}\,\Big[R+d_{1}\,\kappa_{\rm 4D}^{2}\,R_{ab}{}^{cd}R_{cd}{}^{ef}R_{ef}{}^{ab}+d_{2}\,\kappa_{\rm 4D}^{2}\,R_{ab}{}^{cd}R_{cd}{}^{ef}\tilde{R}_{ef}{}^{ab}+d_{3}\,\kappa_{\rm 4D}^{4}\,\mathcal{C}^{2}+d_{4}\,\kappa_{\rm 4D}^{4}\,\tilde{\mathcal{C}}^{2}+d_{5}\,\kappa_{\rm 4D}^{4}\,\mathcal{C}\,\tilde{\mathcal{C}}\Big]\,. (1) Here \kappa_{\rm 4D}^{2}=8\pi G_{\rm 4D} , with G_{\rm 4D} Newton’s constant. The dual Riemann tensor is defined as \tilde{R}_{abcd}=\varepsilon_{abef}R^{ef}{}_{cd} , where \varepsilon_{abcd} is the volume form. We also define the Kretschmann scalar \mathcal{C}=R_{abcd}R^{abcd} and its dual \tilde{\mathcal{C}}=R_{abcd}\tilde{R}^{abcd} . The equations of motion derived from this action are presented in the Supplemental Material (cf. Eq. 7 ). The coefficients d_{K} are Wilsonian parameters of energy dimension -2 ; that is, in a tree-level completion, d_{k}\sim 1/\Lambda^{2} , where \Lambda is the energy scale of new physics. The terms proportional to d_{2} and d_{5} are parity odd, whereas the remaining terms preserve parity. Methodology. We seek the EFT corrections to the QNM spectrum of Kerr black holes. This requires, first, determining the EFT corrections to the Kerr background itself. Because the higher-derivative terms in Eq. ( 1 ) render Kerr no longer an exact solution, we treat the coefficients d_{i} as infinitesimal, so that the EFT-corrected geometry is perturbatively close to Kerr. Perturbation theory then suffices to compute the leading corrections. This strategy has been successfully applied in related contexts 57 ; 45 ; 34 , where EFT-corrected Kerr and Kerr–Newman solutions were obtained. In this first step, we search for stationary and axisymmetric black hole solutions, characterised by a stationary Killing vector \partial_{t} and an axisymmetric Killing vector \partial_{\phi} with \phi\sim\phi+2\pi . A convenient parametrisation is \displaystyle\mathrm{d}s^{2}_{\text{EFTK}} \displaystyle=-\frac{\Delta_{\rm K}(r)}{\Sigma(r,x)}\,F_{1}(r,x)\,\big[\mathrm{d}t-(1-x^{2})F_{4}(r,x)\,\mathrm{d}\phi\big]^{2} (2) \displaystyle+\frac{1-x^{2}}{\Sigma(r,x)}\,F_{3}(r,x)\,\big[F_{4}(r,x)\,\mathrm{d}t-(r^{2}+a_{\rm K}^{2})\,\mathrm{d}\phi\big]^{2} \displaystyle+\Sigma(r,x)\,F_{2}(r,x)\left[\frac{\mathrm{d}r^{2}}{\Delta_{\rm K}(r)}+\frac{\mathrm{d}x^{2}}{1-x^{2}}\right]\,. Here \Delta_{\rm K}(r)=r^{2}-2M_{\rm K}r+a_{\rm K}^{2} and \Sigma(r,x)=r^{2}+a_{\rm K}^{2}x^{2} , with |a_{\rm K}|\leq M_{\rm K} . The metric reduces to Kerr when F_{1,2,3}(r,x)=F^{\rm K}_{1,2,3}=1 and F_{4}(r,x)=F^{\rm K}_{4}=a_{\rm K} , in which case the spacetime coincides with a Kerr black hole of energy E_{K}=8\pi M_{\rm K}/\kappa_{\rm 4D}^{2} and angular momentum J_{K}=a_{\rm K}\,E_{K} . Accordingly, we expand F_{I}(r,x)=F_{I}^{\rm K}+\sum_{k=1}^{2}\frac{\kappa_{\rm 4D}^{2}}{M_{\rm K}^{4}}\,d_{k}\,f_{I}^{(k)}(r,x)+\sum_{k=3}^{5}\frac{\kappa_{\rm 4D}^{4}}{M_{\rm K}^{6}}\,d_{k}\,f_{I}^{(k)}(r,x)\,, (3) for I=1,\ldots,4 , and solve perturbatively for the functions f_{I}^{(k)} (the integration procedure follows 45 ). In the numerical procedure described in 45 , the black hole horizon, where \Delta_{K}(r_{+})=0 vanishes, is held fixed under the EFT deformation, as is the periodicity of \phi . By imposing suitable boundary conditions at asymptotic infinity, we likewise keep the ratio J/E^{2}=J_{K}/E^{2}_{K} fixed. All remaining thermodynamic quantities receive EFT corrections, and we reproduce the results of 57 . The odd-parity terms, proportional to d_{2} and d_{5} , produce no change in any thermodynamic quantity by symmetry, although they do generate non-trivial functions f_{I}(r,x) . Once these corrections are obtained, we proceed to the second step and compute the corresponding EFT‑deformed QNM spectrum. Because the EFT‑corrected black holes are stationary and axisymmetric, all perturbations can be decomposed into Fourier modes of the form e^{-{\rm i}\omega t+{\rm i}m\phi} with m\in\mathbb{Z} . Imposing outgoing boundary conditions at asymptotic infinity and ingoing boundary conditions at the horizon—the defining boundary conditions for QNMs—turns the problem into a non‑Hermitian quadratic Sturm–Liouville eigenvalue problem, whose allowed frequencies \omega are discrete. For Kerr black holes, perturbations separate into radial and polar sectors, and the resulting spectrum can be organised by counting the nodes of the corresponding eigenfunctions. Along the polar direction, the number of zeros is \ell-|m| , implying \ell\geq|m| . Along the radial direction, the number of nodes defines the overtone index n=0,1,\ldots . The QNM spectrum may therefore be labelled as \omega_{\ell mn} . Under the map \phi\mapsto-\phi , the Kerr metric of angular momentum J_{K} is pulled back to a metric in the Kerr family with angular momentum -J_{K} . Under this transformation, m\mapsto-m , so we use this to restrict to J_{K}\geq 0 while keeping m\in\mathbb{Z} . For all Kerr positive spins, the slowest‑decaying mode lies in the sector \ell=m=2 24 ; 47 . This is the sector we focus on when computing EFT corrections, which we define as \omega_{\ell mn}=\omega^{\rm K}_{\ell mn}+\sum_{k=1}^{2}\frac{\kappa_{\rm 4D}^{2}}{M_{\rm K}^{4}}\,d_{k}\,\delta\omega_{\ell mn}^{(k)}+\sum_{k=3}^{5}\frac{\kappa_{\rm 4D}^{4}}{M_{\rm K}^{6}}\,d_{k}\,\delta\omega_{\ell mn}^{(k)}\,, (4) with \{\omega^{\rm K}_{\ell mn}\} denoting the Kerr QNM spectrum. We determine the frequency shifts \delta\omega_{\ell mn}^{\,(k)} through a sequence of steps, formalised in the Supplemental Material, following a variant of the method introduced in 19 —there applied only perturbatively in the spin, with the EFT limit obtained by considering finite d_{k} and subsequently extrapolating to d_{k}\to 0 . In our approach, the d_{k} are taken perturbatively small from the outset (in the true spirit of EFT), while the spin is allowed to take any value consistent with the Kerr extremality bound, |J_{\rm K}|/E_{\rm K}^{2}\leq\kappa_{\rm 4D}^{2}/(8\pi) . The main idea is as follows: One may write down a universal Teukolsky equation for any spacetime, even in the presence of arbitrary matter sources. In general, however, these equations cannot be solved in any practical way for a generic background or for generic forms of matter. The situation simplifies once the equations are linearised around an algebraically special spacetime: in that case one obtains two partial differential equations for two Newman–Penrose components of the Weyl tensor, each sourced by perturbations of the corresponding stress-energy tensor. For typical matter fields, the source terms themselves depend on metric perturbations in a complicated way, so the resulting system is not closed and therefore not solvable by standard decoupling methods. The notable exceptions are cases where the matter sector is sufficiently simple for a closed system to exist (for instance, in Kerr–Newman–(A)dS backgrounds). Our case presents two complications. First, the EFT‑corrected black holes are not algebraically special. Second, the source terms in the Teukolsky equations arise from the effective stress-energy tensor generated by the higher‑derivative corrections. The lack of algebraic speciality is not fatal: our backgrounds differ from Kerr only perturbatively in the small parameters d_{k} , so the departure from type D is itself of order d_{k} (possibly multiplied by Kerr perturbations). This deviation can therefore be treated as an additional source for the Teukolsky equation. The second complication can be handled in much the same way. The effective stress-energy tensor entering the universal Teukolsky equation contributes only perturbatively and depends solely on perturbations of the background geometry—that is, on perturbations of Kerr itself. These effective sources can be constructed using the procedures of 63 ; 27 ; 28 ; 26 , which express metric perturbations of Kerr (and more generally of any type D spacetime) in terms of a pair of complex scalar Hertz potentials, \Psi^{H}_{\pm} , each of which satisfies a sourceless Teukolsky‑type equation. Furthermore, 63 showed that one needs only \Psi^{H}_{-} or \Psi^{H}_{+} to generate the most general metric perturbation of a Kerr black hole (up to trivial shifts in the mass and angular momentum). This is consistent with the expected counting of degrees of freedom: the Hertz potentials are complex scalars, whereas a four‑dimensional graviton carries two real propagating degrees of freedom. The familiar isospectrality of Kerr black holes can then be understood as the statement that, for Kerr, the real and imaginary parts of \Psi^{H}_{-} (or equivalently of \Psi^{H}_{+} ) obey the same QNM spectrum. In practice, however, working directly with the real and imaginary parts of the Hertz potentials leads to cumbersome expressions. It is therefore more convenient to work with perturbations generated simultaneously by \Psi^{H}_{-} and \Psi^{H}_{+} —which naively appears to double the number of degrees of freedom—while imposing the appropriate relations between their real and imaginary parts so that the physical content remains unchanged. We are therefore left with a system of two unknown fields, \psi_{0} and \psi_{4} , obeying Teukolsky equations sourced by an effective (albeit intricate) stress-energy tensor constructed from the Hertz potentials \Psi^{H}_{\pm} . This coupled system can be solved systematically within degenerate perturbation theory. After performing the perturbative expansion, the problem reduces to determining the QNM frequency shifts \delta\omega_{\ell mn} and the mixing between the \Psi^{H}_{-} and \Psi^{H}_{+} sectors implied by the degeneracy structure. Because the degeneracy is two‑fold, each (\ell,m,n) mode yields two distinct corrections, which we denote by \delta\omega^{(k)\;\pm}_{\ell mn} . The numerical procedure used to compute \delta\omega^{(k)\;\pm}_{\ell mn} is described in detail in the Supplemental Material. Results. As noted above, we focus on the EFT corrections to the least‑damped Kerr quasinormal mode, namely the \ell=m=2 , n=0 mode, for all spins satisfying 2 2 2 Our numerical methods can also handle the case j=j_{\rm ext} , but we leave this analysis for future work. j\equiv|J_{\rm K}|/E_{\rm K}^{2}<\kappa_{\rm 4D}^{2}/(8\pi)\equiv j_{\rm ext} . We also note that the Kerr black hole admits a \mathbb{Z}_{2} symmetry, x\to-x , under which the odd‑parity corrections satisfy a characteristic relation (proved in the Supplemental Material), namely \delta\omega^{(k)\;-}_{\ell mn}=-\delta\omega^{(k)\;+}_{\ell mn} (with k=2,5 ). Consequently, for odd‑parity deformations we display only \delta\omega^{(k)\;+}_{\ell mn} in our figures. We compute the EFT‑deformed QNM frequencies at fixed energy and angular momentum; that is, we compare the QNM spectra of the Kerr black hole and its EFT‑deformed counterpart while holding the conserved charges fixed in the microcanonical ensemble. As emphasised in 57 , this ensemble is not well suited for analysing the approach to extremality, owing to the non‑analytic behaviour of certain thermodynamic quantities of Kerr. It is, nevertheless, the physically relevant choice for astrophysical applications of our results. Before presenting the numerical results, we note that the Wilsonian paradigm is most naturally formulated in the Euclidean picture, because integrating out UV modes and defining the RG flow is most conveniently implemented using a Euclidean path integral with weight e^{-S_{\rm E}} . This structure is characteristic of the canonical and grand‑canonical ensembles, which in turn suggests that pathologies appearing in the canonical ensemble may also affect perturbative constructions in other ensembles. In particular, in the canonical ensemble the QNM shifts necessarily diverge when the specific heat of Kerr changes sign. The issue arises because enforcing fixed (T_{H},J_{H}) requires perturbatively adjusting (a_{\rm K},M_{\rm K}) , which introduces the inverse Jacobian \partial(T_{H},J_{H})/\partial(a_{\rm K},M_{\rm K}) . This Jacobian becomes singular when the specific heat at fixed J changes sign 53 , at j/j_{\rm ext}=\sqrt{2\sqrt{3}-3}\approx 0.681 . Correspondingly, the small‑spin expansion of 22 ; 19 ; 21 ; 20 breaks down beyond this point. Figure 1 shows M_{\rm K}\,{\rm Re}\,\delta\omega^{(1)\;+}_{220} (left) and M_{\rm K}\,{\rm Im}\,\delta\omega^{(1)\;+}_{220} (right) over the range 0\sqrt{2\sqrt{3}-3} . All the data generated in this work is available in the GitHub repository . Figure 1: M_{\rm K}\,{\rm Re}\,\delta\omega^{(1)\;+}_{220} (left) and M_{\rm K}\,{\rm Im}\,\delta\omega^{(1)\;+}_{220} (right) as functions of j/j_{\rm ext} . Black squares show our numerical results; grey discs reproduce the data of 14 , publicly available in the GitHub repository . The vertical dashed line at j/j_{\rm ext}=\sqrt{2\sqrt{3}-3} indicates the critical spin where the small‑spin expansion breaks down. We have also examined the behaviour of the spectrum in the grand‑canonical ensemble, where we compare configurations at fixed Hawking temperature T_{H} and angular velocity \Omega_{H} . The QNM shifts are defined by \varpi_{\ell mn}=\varpi^{\rm K}_{\ell mn}+\sum_{k=1}^{2}\kappa_{\rm 4D}^{2}\,\Omega_{H}^{4}\,d_{k}\,\Delta\varpi_{\ell mn}^{(k)}\\ +\sum_{k=3}^{5}\kappa_{\rm 4D}^{4}\Omega_{H}^{6}\,d_{k}\,\Delta\varpi_{\ell mn}^{(k)}\,, (5) where \varpi_{\ell mn}\equiv\omega_{\ell mn}/T_{H} . For Kerr, the mode \varpi_{220} is known to approach a constant in the extremal limit, implying that \omega_{220} tends to zero as extremality is approached. Our goal is to determine how this behaviour is modified once higher‑derivative corrections are included in the EFT description. The results of 44 strongly suggest that \Delta\varpi_{220}^{(k)} remains finite in the extremal limit. This follows from the fact that the near‑horizon geometries constructed in 44 possess an SO(2,1) symmetry that enforces this behaviour and are completely smooth, even though the full spacetime is only mildly singular. In Fig. 2 we show the real (black disks) and imaginary (grey squares) parts of the QNM EFT shifts in the grand‑canonical ensemble, for the same EFT deformations as in Fig. 1 , plotted as a function of \tau_{H}\equiv T_{H}/\Omega_{H} on a log–log scale. Although normalising the EFT shifts by \Omega_{H} is natural in the grand‑canonical ensemble, it introduces an artificial singular behaviour in the \Omega_{H}\to 0 (or \tau_{H}\to+\infty ) limit, as seen in Fig. 2 . As we approach \tau_{H}=0 , the shifts exhibit an oscillatory pattern with an approximately fixed frequency in \log\tau_{H} . The remaining deformations display the same qualitative behaviour, although for the k=3,4 sectors the oscillations only become visible at lower values of \tau_{H} . The “echoing period” of the real part is approximately twice that of the imaginary part across all EFT sectors we have investigated. Figure 2: Real (black disks) and imaginary (grey squares) parts of the QNM EFT shifts in the grand‑canonical ensemble versus \tau_{H}\equiv T_{H}/\Omega_{H} (log–log scale). The curves develop logarithmic oscillations toward \tau_{H}\to 0 , with the real‑part “echoing period” roughly twice that of the imaginary part. Discussion. We have computed the EFT corrections to the QNM spectrum of rotating Kerr black holes for the first five non-trivial higher-derivative operators, working perturbatively in the Wilson coefficients d_{k} but non-perturbatively in the dimensionless spin j/j_{\rm ext} . In all cases, we find that the small-spin expansion breaks down well before the extremal regime, indicating that perturbation theory around slowly rotating solutions has a parametrically limited domain of validity. This breakdown reflects the limitations of the auxiliary expansion in spin rather than any failure of the EFT itself. By contrast, our computation remains uniformly controlled for all 0\leq j