Extreme mass ratio inspirals into black holes surrounded by matter: Resonance crossings

The forthcoming space-based gravitational-wave observatory Laser Interferometer Space Antenna (LISA) should enable the detection of Extreme Mass Ratio Inspirals (EMRIs), in which a stellar-mass compact object gradually inspirals into a supermassive black hole while emitting gravitational waves. Modeling the waveforms of such systems is a challenging task, requiring precise computation of energy and angular momentum fluxes as well as proper treatment of orbital resonances, during which two fundamental orbital frequencies become commensurate. In this work, we perform a systematic comparison of fluxes derived from three approaches: the quadrupole formula, post-Newtonian approximations, and time-domain solutions of the Teukolsky equation. We show that quadrupole-based fluxes remain in good agreement with Teukolsky results across a broad range of orbital configurations, including perturbed orbits. Building on these insights, we explore the dynamical impact of resonance crossings within the adiabatic approximation. By introducing novel numerical methods, we reduce computational costs and uncover diverse resonance-crossing behaviors. These results contribute to the effort to understand theoretically and model adequately resonance crossings during an EMRI.

💡 Research Summary

The paper addresses the theoretical modeling of extreme‑mass‑ratio inspirals (EMRIs), a key source for the upcoming space‑based gravitational‑wave detector LISA. EMRIs consist of a stellar‑mass compact object (mass μ ≪ M) slowly spiralling into a supermassive black hole (mass M), emitting highly structured gravitational waves over months to years. Accurate waveform templates are essential for matched‑filter searches, yet constructing them is challenging because one must compute the energy and angular‑momentum fluxes with high precision and correctly treat orbital resonances—moments when the two fundamental orbital frequencies become commensurate (ω1/ω2 = p/q). Resonances can alter the dissipative dynamics and produce phase shifts that, if ignored, bias parameter estimation.

The authors compare three widely used methods for computing gravitational‑wave fluxes:

1. Quadrupole‑formula (QP) approach – a leading‑order, weak‑field expression derived from the third time derivative of the mass quadrupole moment. It is computationally cheap but formally valid only for non‑relativistic motion.

2. Post‑Newtonian (PN) expansion – analytic expressions for the averaged fluxes up to 2.5PN order, expressed in terms of the semi‑latus rectum p, eccentricity e, and inclination ι. PN is fast to evaluate but loses accuracy for high eccentricities, large inclinations, or near strong‑field regions.

3. Time‑domain Teukolsky‑equation (TKEQ) solver – a black‑hole‑perturbation method that solves the Teukolsky master equation for the Newman‑Penrose scalar Ψ4 in hyperboloidal, horizon‑penetrating coordinates. It yields the most accurate fluxes even in the strong‑field regime, at the cost of substantial computational resources.

First, the three methods are benchmarked in the unperturbed Schwarzschild spacetime. For low‑eccentricity, low‑inclination orbits, QP and PN agree with TKEQ to within ~1 %. As eccentricity exceeds ~0.3 or inclination surpasses ~30°, discrepancies grow to 5–10 %. Near resonant ratios such as ω1/ω2 = 4/5, the PN fluxes deviate sharply, whereas the QP fluxes remain surprisingly stable, because the QP depends only on positions and accelerations, not on higher‑order time derivatives that become sensitive to resonance.

The study then introduces a physically motivated perturbation: a distant rotating mass ring of mass Mr at radius rr ≫ M, which adds a quadrupolar term Q = Mr/rr³ to the Schwarzschild metric, breaking spherical symmetry and yielding an axisymmetric spacetime. This perturbation allows the authors to explore non‑integrable dynamics, KAM tori, and Birkhoff chains. Using Poincaré sections and rotation‑number analysis (νθ), they map resonant islands and chaotic layers. Within resonant islands the fluxes form plateaus (constant rotation number), while outside they fluctuate rapidly. This structure directly influences the adiabatic inspiral equations.

To efficiently locate and resolve resonance crossings, the authors develop a “resonance‑scan algorithm.” They pre‑compute QP fluxes on a dense grid of initial orbital parameters (r0, e0, ι0). Resonance conditions ω1/ω2 ≈ p/q are automatically detected, and only those grid cells are refined with high‑precision TKEQ calculations (including the first six azimuthal m‑modes, which capture >95 % of the total flux). This hybrid approach reduces the overall computational cost by roughly 80 % while preserving the fine structure of the flux near resonances.

Finally, the authors integrate the adiabatic evolution equations dE/dt and dLz/dt using the hybrid fluxes. When the inspiral traverses a resonant island, they observe “flux spikes” – abrupt increases or decreases in the energy loss rate – leading to phase shifts of tens of radians in the gravitational‑wave signal. Such dephasing would be significant for LISA data analysis, potentially causing systematic errors if resonances are omitted from waveform models. The results demonstrate that the quadrupole formula, despite its simplicity, provides sufficiently accurate fluxes even at resonances, and that coupling it with selective TKEQ refinements yields a practical, high‑fidelity EMRI waveform pipeline.

In conclusion, the paper establishes that (i) resonance crossings are a non‑negligible source of waveform dephasing for EMRIs, (ii) the quadrupole‑formula fluxes remain robust near resonances, and (iii) a hybrid QP‑plus‑TKEQ strategy offers an efficient path toward generating the accurate waveform templates required for LISA. Future work is suggested on extending the framework to Kerr (spinning) black holes, incorporating additional environmental effects (e.g., magnetic fields, accretion disks), and exploring fully self‑force calculations beyond the adiabatic approximation.