Second-Order Self-Force and EMRI Accuracy

The question is not whether second-order gravitational self-force theory exists, but whether it already provides a completed, independently validated waveform for a generic extreme-mass-ratio inspiral (EMRI). At the evidence cutoff of 2026-08-20, the answer is mixed in a useful way: first-postadiabatic waveforms are operational and scientifically informative for restricted quasicircular systems, while eccentric, inclined, rapidly spinning Kerr inspirals still lack the complete second-order inputs, resonance and plunge treatment, and end-to-end LISA error budget needed for a generic accuracy claim.

Helpful background. Second-Order Self-Force and Nonlinear Regularization develops the perturbative field equations, puncture construction, regularization, and gauge qualifications used here. This page assesses the present reach of that machinery rather than repeating its derivation.

The finite EMRI accuracy question

Write the mass ratio as $\varepsilon=m/M\ll1$. An adiabatic, or 0PA, model captures the leading slow drift through geodesics. A first-postadiabatic, or 1PA, waveform additionally needs specific first- and second-order self-force information; secondary spin and transition effects enter at comparable bookkeeping orders in relevant formulations. “Second order is available” is therefore meaningful only after specifying the orbit family, background spin, retained terms, observable, evolution interval, and accuracy norm.

Four axes control this assessment. First is perturbative content: a second-order local field, a dissipative flux, and a fully evolved 1PA waveform are different deliverables. Second is orbital scope: circular Schwarzschild is not generic eccentric-inclined Kerr. Third is the error norm: residuals in an effective source do not directly bound accumulated phase, mismatch, or parameter bias. Fourth is validation: internal convergence, balance laws, PN overlap, comparison with EOB or numerical relativity, and an independent implementation answer different questions.

Assessment at 2026-08-20	Meaning
Question status	Partially resolved
Evidence status	Supported
Confidence	High for the stated restricted success and for the identified missing generic ingredients
Domain	1PA gravitational waveforms demonstrated primarily for quasicircular, nonspinning or slowly spinning systems; not generic eccentric-inclined Kerr through resonances and plunge
What could change the conclusion	An independently benchmarked generic-orbit waveform with a closed error budget appears

Entry project: locate the reliable part of one circular inspiral

Choose this route if you enjoy perturbation theory and numerical diagnostics. You need orbital dynamics, asymptotic expansions, and enough Mathematica to inspect interpolation and differentiate a phase. A small waveform study is suitable for a laptop with access to the required kernel; producing new second-order Kerr field data requires specialist supervision and substantially different computing resources.

Start with Wardell et al. 2023, waveform construction and comparison figures, extracting the orbit restrictions and terms included at 1PA. Then follow the Black Hole Perturbation Toolkit, WaSABI workflow example: query InitialConditions, ListModes, and CiteModel for 1PAT1, and record the exact code revision before generating a waveform. Select one nonspinning circular comparison from Albertini et al. 2022, §§ III–V; identify its mass convention, frequency interval, comparator, and alignment before attempting its diagnostic. Do not substitute an arbitrary mass ratio and call the resulting plot a reproduction.

The finite deliverable is a notebook for one configuration containing the phase difference or $Q_\omega=\omega^2/\dot\omega$, where $\omega$ is the instantaneous waveform angular frequency and the dot denotes a time derivative, with separate changes under interpolation refinement, phase differentiation, and moving the inspiral endpoint. Include a table of omitted physical terms. Reproducing the software example is training; matching the published comparison additionally requires its input waveform and conventions.

A candidate extension, with novelty unchecked, is to determine which of those numerical choices limits the diagnostic over a small, fixed circular parameter set. Stop the extension if numerical differentiation or endpoint choice dominates the supposed physical discrepancy. If the comparator cannot be obtained, pivot to a documented internal convergence study and keep the claim at that level. Discuss gauge-invariant interpretation with a self-force researcher before attributing disagreement to missing second-order physics.

What restricted second-order calculations establish

The formal programme is on firm ground. Pound gives a self-consistent second-order equation of motion and a practical puncture-based route to regular fields, while emphasizing the gauge and matching structure required for a controlled expansion Pound 2017, §§ II–V. That is a formulation result, not by itself a mission-ready waveform.

The decisive computational advance is narrower and stronger. Wardell and collaborators used complete second-order information to construct fast 1PA waveforms for nonspinning, quasicircular binaries; their phase comparisons show that second-order information changes the waveform materially rather than serving as a formal refinement Wardell et al. 2023, main text and waveform-comparison figures. Upton and collaborators document the effective source behind this production calculation, including quadratic mode coupling, puncture structure, and slow-evolution checks Upton et al. 2026, §§ II–VI and appendices.

The domain is expanding. A restricted model including a slowly spinning primary and arbitrary secondary spin has been compared with numerical relativity for roughly $q\gtrsim5$, but it remains quasicircular and assumes small primary spin and misalignment Mathews et al. 2026, model construction, domain statement, and NR comparisons. Separately, the architecture for generic eccentric and precessing Kerr evolution is known: it identifies which averaged forces, spin couplings, and two-timescale ingredients a complete calculation must supply Mathews and Pound 2025, §§ II–VI. Architecture is valuable because it makes the missing inputs finite; it does not make those inputs numerically available.

Results and open questions at 2026-08-20

Claim or evidence	What it supports	What it does not establish
Second-order puncture and equation-of-motion framework	A mathematically controlled route to 2SF dynamics	A computed generic Kerr inspiral
Quasicircular 1PA waveform construction	Operational waveform impact in circular Schwarzschild	Eccentricity, inclination, rapid primary spin, resonances, or plunge
Effective-source tests	Internal analytic and numerical consistency of the production source	Independent end-to-end reproduction
Slowly spinning extension and NR comparison	Useful reach beyond the strict EMRI/nonspinning corner	Generic Kerr or LISA-duration accuracy
Generic two-timescale architecture	A finite inventory of required ingredients	Availability of the generic 2SF data
1PAT1–EOB–NR comparison	An external diagnostic of phase evolution and endpoint error	A universal error bound for all mass ratios and orbit families

The rows are not votes. The quasicircular waveform and source papers carry the positive claim; the architecture and review delimit what remains; the independent comparison tests only a restricted sector. Together they justify “partially resolved,” not “almost complete.”

Strongest competing interpretations

The optimistic interpretation is that the hardest conceptual and computational obstacles have been crossed: a genuine 2SF source has produced efficient 1PA waveforms, several independent limits agree, and genericisation is now an engineering sequence. This is credible within the circular programme.

The cautious interpretation is that generic Kerr changes the character of the problem. Eccentric and inclined motion introduces multiple frequencies, turning points, resonance crossings, a larger mode-coupling problem, and more demanding long-time interpolation. A method that is accurate away from the last stable orbit can also fail during transition. The LISA Waveform Working Group accordingly treats 1PA inputs as essential while still identifying generic implementation and validation as incomplete LISA Waveform Working Group 2025, EMRI/self-force and accuracy-requirement sections. The evidence favors neither “mere engineering” nor “fundamental obstruction”; it establishes a functioning restricted programme with a nontrivial remaining scope.

Counterexamples, null results, and failed routes

No cited result refutes second-order self-force theory. The strongest adverse evidence instead blocks overgeneralization. Albertini and collaborators compare 1PAT1 with TEOBResumS and numerical relativity using $Q_\omega$ and dephasing diagnostics. They find useful agreement but also mass-ratio-dependent discrepancies and significant contamination near transition to plunge Albertini et al. 2022, §§ III–V. A waveform can therefore be correct through much of an inspiral yet lack a closed full-signal error budget.

The most informative null search is bibliographic and technical: no archival result found by the cutoff combines eccentric and inclined Kerr motion, complete needed 2SF information, transient-resonance handling, transition and plunge, independent implementation, and a LISA-duration phase-and-amplitude budget. This absence does not show the goal is unattainable. It shows that a generic-completion claim has not yet been demonstrated.

From a regular field to a mission-scale statement

Several error layers must be propagated rather than collapsed. Discretization and mode-sum truncation affect the local field; interpolation affects slowly evolving orbital data; omitted perturbative orders accumulate over $O(\varepsilon^{-1})$ cycles; resonance and endpoint models can produce localized phase changes; and detector response converts waveform error into parameter bias. A residual plot for one orbit cannot certify all these layers.

The appropriate target is a complete error budget expressed in physical outputs: phase and amplitude by harmonic, detector-weighted mismatch over a declared observation time, and bias relative to posterior width for representative LISA signals. The same waveform should pass balance laws and invariant overlap checks before its detector-level performance is interpreted. This is why public software is necessary but not sufficient evidence.

What would change the assessment

The assessment would move toward resolved if an independent team produced a generic eccentric-inclined Kerr 1PA waveform, including the required second-order data, secondary spin, resonances, and a controlled transition, then demonstrated convergence and agreement in gauge-invariant observables against another implementation. The release would need a closed error budget propagated to LISA phase, mode amplitudes, mismatch, and parameter bias over representative mission durations.

Conversely, a persistent invariant disagreement between independent 2SF implementations, failure of balance laws after numerical convergence, or a demonstrated order-unity phase error in the nominal domain would lower support. A new circular waveform alone would improve precision without resolving the generic question.

Research checklist

• Choose one orbit family and state $q$, primary and secondary spins, eccentricity, inclination, initial frequencies, observation duration, and endpoint.
• Identify every included 0PA and 1PA term; list resonance, memory, absorption, transition, and plunge terms as included, approximated, or absent.
• Pin the WaSABI commit or immutable release, Mathematica/kernel version, model name, interpolation data, harmonic set, and numerical tolerances.
• Reproduce a published phase or $Q_\omega$ comparison before extending the domain; record alignment and time/frequency conventions.
• Test resolution, multipole truncation, interpolation, balance laws, and at least one PN/EOB/NR overlap invariant separately.
• Propagate the numerical and perturbative errors to phase, amplitude, detector-weighted mismatch, and parameter bias.
• Label a completed rerun as such. If only a protocol was prepared, do not describe it as reproduction.

Source and data map

Pound supplies foundational analytic formulation. Wardell et al. and Upton et al. provide the primary quasicircular waveform and effective-source results. Mathews and collaborators supply the spin extension and generic architecture. Albertini et al. provide the strongest partially independent waveform comparison, while the LISA Waveform Working Group provides current requirements context. The Black Hole Perturbation Toolkit WaSABI documentation exposes restricted 1PA and adiabatic Kerr models, but the live page did not expose a unique immutable release during this check.

The reproducible boundary is therefore a declared restricted-model rerun, not a generic Kerr verification. Public code can generate a named 1PAT1-family waveform and support comparison with an identified EOB or SXS waveform. It cannot reproduce generic 2SF data that have not been released, and the exact WaSABI commit or archive must be supplied by the researcher. No independent end-to-end reproduction was performed for this assessment; the checklist is a reproduction prescription.

References

• Albertini, A., Nagar, A., Albanesi, S., Bernuzzi, S., Damour, T., and Rettegno, P. “Comparing second-order gravitational self-force, numerical relativity and effective one body waveforms from inspiralling, quasi-circular and nonspinning black hole binaries.” Physical Review D 106, 084061 (2022). Open preprint.
• Black Hole Perturbation Toolkit. “WaSABI.” Live model documentation, checked 20 August 2026. Project documentation.
• LISA Waveform Working Group. “Waveform modelling for the Laser Interferometer Space Antenna.” Living Reviews in Relativity (2025). DOI.
• Mathews, J. and Pound, A. “Post-adiabatic waveform-generation framework for asymmetric precessing binaries.” Physical Review D 112, 104078 (2025). DOI; open preprint v3.
• Mathews, J., et al. “Post-adiabatic self-force waveforms: slowly spinning primary and precessing secondary.” Physical Review D 113, 064034 (2026). DOI; open preprint v2.
• Pound, A. “Nonlinear gravitational self-force: second-order equation of motion.” Physical Review D 95, 104056 (2017). DOI.
• Upton, S. D., et al. “Effective source for second-order self-force calculations: Quasicircular orbits in Schwarzschild spacetime.” Physical Review D 113, 064013 (2026). DOI; open preprint v2.
• Wardell, B., et al. “Gravitational waveforms for compact binaries from second-order self-force theory.” Physical Review Letters 130, 241402 (2023). DOI.