{"id":"ef091328-9251-4823-b941-15b7760da511","arxiv_id":"2411.09147","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A 12PN analytic phase model for quasi-spherical inclined EMRIs in Kerr spacetime is presented, and TaylorT1 is found to converge fastest among the time-domain approximants.","lead":"Researchers derived 12th-order post-Newtonian formulas for the energy and angular momentum loss of a small object on a spherical inclined orbit around a spinning black hole, and built gravitational-wave phase templates from them. The templates give LISA-style searches an analytic description of extreme mass-ratio inspirals, with the TaylorT1 approximant showing the best internal convergence.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 12PN flux coefficients that drive every claimed convergence result are not in the paper, only on the BHPC website [31]; the 'TaylorT1 best' ranking and the 8PN/9PN accuracy thresholds rest on those unseen numbers.","rationale":"The paper is a serious and largely clear construction: it adapts the standard TaylorT/TaylorF formalism to spherical Kerr inspirals, uses a reasonable adiabatic first-order dissipative self-force framework, and provides explicit coefficients through 4.5PN. The mathematical structure of the template families is plausible, and the convergence diagnostics are internally coherent. The load-bearing weakness is exactly what the reader identified: the 12PN coefficients, which are the paper's chief new input and which control the claimed accuracy thresholds and the ranking of approximants, are not present in the manuscript itself. The text states that these coefficients were obtained and validated against numerical data, but supplies no derivation, no tables for k>9, and only a website reference. If the online coefficients contain an error, then the 8PN/9PN dephasing statements, the TaylorF2 mismatch results, and the 'TaylorT1 is best' conclusion are all unsupported. The extra internal inconsistency between Eq. (45) and Eq. (26) is a concrete reminder that typographical or coefficient-level slips can occur in exactly the places that would matter for such an external check. I did not find a flaw in the adiabatic reduction or in the Taylor-family derivation that would by itself invalidate the method, and the concurrent independent work [34] provides a feasible way to settle the concern. Since the reader already assigned CONDITIONAL with the same central worry, my read does not change the verdict; it strengthens the case for requiring the external coefficient check before full acceptance.","tokens_in":32597,"tokens_out":10600,"duration_ms":117795,"concrete_test":"Retrieve the BHPC coefficient files [31] as referenced (ideally archiving the exact version), independently re-expand \\dot E and \\dot L to 12PN, and recompute the TaylorT1/TaylorT2 dephasing \\Delta\\Phi^{(24)} for System1 with q=0.9 and Y_I=0.9, as well as the TaylorF2 mismatch in Fig. 12. Then compare the k=10..24 coefficients against the independent 12PN results of Ref. [34] and the numerical fluxes of Refs. [14,15] on the same (q,Y,x) grid. If any coefficient differs, or if the recomputed 8PN dephasing no longer stays below 0.1 for TaylorT1 or the 12PN mismatch changes materially, the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central assertion of Sec. II B and Sec. IV is that the k<=24 (12PN) coefficients of \\dot E and \\dot L have been obtained, agree with the numerical flux data of Refs. [14,15] to numerical precision, and produce TaylorT1 templates that keep dephasing below O(0.1) for System1 at 8PN. Every one of these statements is evaluated using coefficients k=10 through 24 that are not derived or displayed anywhere in the manuscript; Appendices A-I stop at k<=9 (4.5PN) and refer to the Black Hole Perturbation Club website [31] for the higher terms. The manuscript is therefore not self-contained at the claimed order: a reader cannot check the 10PN/12PN curves in Figs. 1-12, the phase differences in Tables I-II, or the ranking of Taylor families without downloading an unversioned external file. A concrete warning that such checks matter is the internal inconsistency between Eq. (45) and Eq. (26), which the text says are equivalent after x -> x_f: the 4PN coefficient in Eq. (45) is -80209/8064, while Eq. (26) gives -81217/8064 - [(33+15Y^2)/128] q^2. This does not by itself overturn the headline, but it shows that displayed coefficients can contain slips; if a similar slip exists in the high-order website coefficients, the claimed 12PN templates and accuracy thresholds would shift. The independent 12PN flux calculation [34] appearing simultaneously is a ready cross-check, but the present manuscript does not use it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs post-Newtonian (PN) template families for the gravitational-wave phase of a compact object on a quasi-spherical (e = 0, inclined) orbit in Kerr spacetime, using 12PN analytic formulas for the orbital energy, angular momentum, and their orbit-averaged rates of change. The authors build TaylorT1, TaylorT2, TaylorT3, TaylorT4, alternative T1a/T4a, and TaylorF2 approximants and assess convergence by the size of the last included PN correction: the dephasing between successive truncations for the TaylorT families and the mismatch between successive TaylorF2 truncations. They report that TaylorT1 (and T1a) converge fastest, TaylorT2 next, and TaylorF2 comparably to TaylorT2, and that for a reference 'System1' EMRI the 8PN TaylorT1 keeps the dephasing below O(0.1) over two years, whereas for the late-inspiral 'System2' even 12PN is insufficient. Only coefficients up to k <= 9 (4.5PN) are printed in Appendices A-I; the k = 10-24 coefficients, which drive all the 10PN/12PN results, are hosted on the unversioned Black Hole Perturbation Club website (Ref. [31]). No parameters are fitted to data.","tokens_in":32851,"tokens_out":17698,"duration_ms":165557,"significance":"If the unpublished k <= 24 coefficients are correct, this is a valuable step toward accurate EMRI template banks for LISA: it is, to my knowledge, the first systematic high-order PN template construction for inclined spherical orbits in Kerr, and the fully analytic T2/F2 forms are well suited for fast searches and for parameterized beyond-GR extensions. The paper's strengths include the explicit display of a large set of coefficients through 4.5PN for the energy, fluxes, Y(x), and each template family; a parameter-free construction with no fitting to data; and a clear statement of the adiabatic, first-order dissipative approximation. Its weaknesses are that the load-bearing high-order coefficients are not in the manuscript, the asserted consistency with numerical data is not shown, one displayed low-order formula is internally inconsistent with another, and the 'best performance' ranking is based on a self-convergence diagnostic rather than an external accuracy reference. The central claims are plausible but cannot currently be verified independently from the paper's contents.","major_comments":[{"comment":"The central quantitative results -- the 12PN convergence curves in Figs. 1-12, the 12PN entries of Tables I-II, and the accuracy thresholds quoted in Sec. IV (e.g., '8PN TaylorT1 keeps the dephasing below O(0.1)') -- depend on the energy and flux coefficients for k = 10 through 24, but these coefficients are not derived, tabulated, or deposited in the manuscript: Appendices A-I stop at k <= 9, and the text refers to the unversioned 'Black Hole Perturbation Club' website (Ref. [31]) for the higher-order terms. A reader therefore cannot check the claimed 12PN accuracy from the paper alone. In addition, the Sec. II B assertion that the 12PN formulas are 'consistent with the numerical results [14,15] within numerical precision' is not backed by any plot, table, or quantitative measure shown in the paper. I ask that the k <= 24 coefficients be supplied in a versioned ancillary file or appendix, and that the comparison with Refs. [14,15] -- and ideally with the independent 12PN calculation of Ref. [34], which is cited but not used -- be presented explicitly.","section":"Sec. II B / Sec. IV / Apps. A-I / Ref. [31]"},{"comment":"The text states that Eq. (45) is equivalent to Eq. (26) with x replaced by x_f, but the displayed dY/dx formulas disagree at order x^4: Eq. (26) gives the coefficient -[81217/8064 + ((33 + 15Y^2)/128) q^2], while Eq. (45) gives -80209/8064 with no q^2 term. The x^5 coefficient shown in Eq. (47), 80209/40320, likewise differs from the 81217/40320 that appears both in the integral of Eq. (26) and in the coefficient tilde-Y5 printed in Appendix D, and the displayed x^4 term in Eq. (47) carries an extra power of q relative to a direct integration of Eq. (26). This is a concrete slip at 2PN order, far below the claimed 12PN reach, and it weakens confidence that the unchecked high-order coefficients are free of similar slips; please reconcile the displayed formulas and re-derive the Appendix G F2 coefficients from a single consistent integration.","section":"Eqs. (26), (45), (47); App. D"},{"comment":"The family ranking and the headline thresholds are obtained from self-convergence diagnostics: Delta-Phi^(n) compares successive truncations of the same family (Eq. (59)), and the TaylorF2 mismatch in Eq. (61) compares adjacent orders of the same approximant. These quantities measure the size of the last included correction, not the distance of any template from the true waveform. Consequently the Abstract's 'best performance' and Sec. V's '8PN TaylorT1 template is expected to keep the dephasing less than O(0.1)' are statements about internal convergence that have not been calibrated against an external reference; the paper reports having the numerical flux data of Refs. [14,15] to hand but shows no such comparison. I recommend either qualifying the accuracy language throughout, or adding one external calibration (e.g., the fluxes of Refs. [14,15] or a self-forced inspiral such as Ref. [35]) to substantiate the claimed accuracy thresholds.","section":"Sec. IV A-C; Eqs. (58)-(61); Abstract; Sec. V"}],"minor_comments":[{"comment":"The paper cites but does not engage with the simultaneous 12PN flux calculation of Ref. [34]; since that work provides a direct check of the k = 10-24 coefficients, a short comparison would materially increase confidence in the headline results.","section":"Footnote 1 / Ref. [34]"},{"comment":"Equation (36) and the surrounding text do not state the order at which the TaylorT3 variable inversion (Eq. (35)) is truncated; please state explicitly that the inversion is truncated at the same PN order as the underlying T2 expansion.","section":"Sec. III D"},{"comment":"The definitions of the polygamma-based functions Psi_A and Psi_B appear only in Appendix B, although the same functions are used in Appendices D-I; add a pointer or repeat the definitions at first use in each appendix.","section":"Apps. B-I"},{"comment":"The 'spike bottoms' attributed to logarithmic terms in Figs. 1-2 are left unexplained; a sentence describing why the log terms produce cusps in Delta-E-dot and Delta-L-dot would help the reader assess the claimed convergence.","section":"Sec. IV A"},{"comment":"Tables I-II list x_fin values 'evaluated by solving the TaylorT1 equations at the 12PN order'; because these values inherit the unverified high-order coefficients, a footnote stating this dependence would be helpful.","section":"Sec. IV, Tables I-II"},{"comment":"The reference line Delta-Phi = 0.05 (rho = 20) is introduced in the caption of Fig. 3 but not in the main text; move or repeat the definition at the first use in Sec. IV B.","section":"Sec. IV B / Fig. 3"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle is archival: the paper's central claims rest on coefficients that live on an unversioned website. If the authors provide the coefficients in a versioned ancillary file and add the numerical comparisons they assert in Sec. II B, I would view the paper as publishable. The Eq. (45)/(26) slip is localized, and the appendices otherwise look carefully assembled, so I do not regard it as evidence of systemic error. I would also ask the editor to have the authors clarify the status of Ref. [34], which appeared simultaneously and whose results can resolve the verification problem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a look if you work on EMRI templates. The genuinely new thing here is a 12PN analytic flux/energy/angular-momentum set for spherical Kerr orbits with arbitrary inclination, plus TaylorT1a/T4a/F2 template families built from the two-parameter (x,Y) dynamics. The k<=9 coefficients in the appendices look consistent with the earlier 4PN/5PN generic-orbit papers, and the template derivation is careful. The convergence study is honest: they compare successive PN orders, find TaylorT1/T1a best, T2/F2 next, T3/T4 poor, and they say plainly that for the late-inspiral System2 even 12PN does not keep dephasing below 0.1 rad over two years. They also flag the e=0 restriction and the neglect of conservative self-force. All of that is solid subfield work and worth publishing.\n\nThe soft spots are real but not fatal. First, the 12PN coefficients (k=10..24) that drive every 8PN/10PN/12PN curve and every ranking are not in the manuscript; they are on an unversioned BHPC website. A referee cannot check the central claim from the paper alone. Second, the agreement with numerical flux data [14,15] is asserted, not shown; no comparison plot or table appears. Third, the template ranking is strictly an internal self-convergence test—each family is compared with its own lower-order truncation—so 'best performance' means best internal convergence, not closeness to an exact waveform. That is a legitimate way to rank PN families, but it is worth saying explicitly. Fourth, there is a displayed inconsistency: Eq. (45) gives the 4PN dY/dx_f coefficient as -80209/8064, while Eq. (26) has -81217/8064 - (33+15Y^2)/128 q^2. The F2 phase coefficients in Appendix G presumably use the correct form, but the mismatch should be fixed in revision. The independent 12PN flux calculation [34] is a ready cross-check and should be cited and commented on in the revised version.\n\nBottom line: the method is established, the low-order coefficients check out, and the template families are a useful resource for LISA studies. As written, the paper is not fully self-contained and has a small internal inconsistency, so it needs revision rather than unconditional acceptance, but it absolutely deserves a serious referee. I would send it out with a request for ancillary coefficient files and at least one numerical cross-check.","headline":"Genuinely useful 12PN template extension for inclined spherical EMRIs, but the central coefficients live on a website and one displayed coefficient is internally inconsistent; referee it with requests for self-contained data.","tokens_in":33448,"tokens_out":3373,"would_cite":true,"duration_ms":122898,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["04.30.-w","04.25.Nx","04.70.Bw"],"model":"deepseek-v4-flash","headline":"The paper extends PN phase templates for spherical EMRI orbits in Kerr spacetime to 12PN order and ranks the TaylorT1 approximant as the most convergent.","keywords":["extreme mass ratio inspiral","post-Newtonian templates","Kerr spacetime","spherical orbits","Taylor approximants","dephasing","gravitational wave phase","black hole perturbation theory"],"falsifier":"Recompute the 12PN flux coefficients for k=10 through 24 using an independent frequency-domain black-hole perturbation code and use them to evolve a two-year TaylorT1 inspiral for q=0.9, YI=0.9 in System1; if the resulting dephasing relative to a fully numerical self-force trajectory exceeds 0.1 rad, or if any coefficient disagrees with the online values, the paper's convergence claim would be overturned.","tokens_in":32345,"feed_emoji":"🛰️","tokens_out":10384,"duration_ms":105354,"temperature":0.7,"pith_summary":"The paper derives post-Newtonian (PN) gravitational-wave phase templates for a compact object on a spherical orbit—constant radius but inclined relative to the black hole's equatorial plane—around a spinning black hole. Using 12PN analytic expressions for the orbital energy, angular momentum, and their radiation-driven rates of change, it constructs several Taylor families of phase approximants and measures how quickly successive PN orders converge. It finds that the TaylorT1 approximant converges fastest, that TaylorT2 and the frequency-domain TaylorF2 are close behind, and that the fully analytic T2/F2 forms remain practical for template banks and for studying deviations from general relativity. This matters for space-based gravitational-wave detectors, where a phase error of order one radian destroys the match between a signal and its template.","feed_headline":"12PN templates rank TaylorT1 best for inclined EMRIs","feed_subtitle":"Fully analytic T2 and F2 forms come close, reaching 0.1-rad accuracy at 9PN for early inspirals.","key_machinery":"The object that carries the argument is the Taylor approximant, built from the adiabatic equations dx/dt and dY/dt obtained by inverting the Jacobian between (E,L) and (x,Y) using the 12PN flux formulas. TaylorT1 leaves the inverse Jacobian unexpanded and integrates numerically; TaylorT4 expands the right-hand side in x before integrating; TaylorT2 and TaylorF2 series-expand and then integrate analytically to give closed-form phase expressions; TaylorT3 rewrites the phase as a function of remaining time. The convergence diagnostics are the order-by-order phase differences $\\Delta$ Phi_X^(n) between successive PN orders and the frequency-domain mismatch M_F2^(n), with reference lines at 0.05 rad and $10^{{-3}}$ corresponding to SNR 20. The flux expansions contain log x terms, Euler's constant, and polygamma-based functions of the black hole spin, and the leading terms in dY/dt cancel so the inclination evolution effectively starts at 1.5PN.","core_discovery":"The central claim is that for quasi-spherical EMRIs in Kerr spacetime, the 12PN adiabatic evolution equations produce usable phase templates, and that among them the TaylorT1 approximant has the best convergence. The paper writes the specific energy E and angular momentum L, together with their averaged dissipative rates, as PN series in x=(M omega_phi)^{1/3} and the inclination variable Y=cos iota, and checks consistency with numerical flux calculations. Solving the unexpanded evolution equations yields TaylorT1; expanding the right-hand sides yields TaylorT4; analytic integration yields TaylorT2, TaylorF2, and TaylorT3, with two alternatives (T1a, T4a) that eliminate Y through its series Y(x). For an early-inspiral system, the 8PN TaylorT1 keeps two-year dephasing below about 0.1 rad, while TaylorT2 needs 9PN for comparable accuracy and TaylorF2 behaves like TaylorT2. For a late-inspiral system, even the 12PN templates cannot hold dephasing below 0.1 rad, and the convergence worsens with larger black hole spin and larger initial inclination.","pith_inferences":["Editorial inference: Because the paper measures convergence by differences between successive PN orders rather than against an exact waveform, the claimed 'best performance' of TaylorT1 is a statement about series convergence; an independent numerical self-forced inspiral would quantify absolute phase error.","Editorial inference: The same adiabatic construction should extend to eccentric orbits, and the paper notes the T1/T4 extension is trivial; if the pattern holds, TaylorT1 would again be the safest time-domain family for generic EMRIs, but the analytic T2/F2 forms would be harder to obtain.","Editorial inference: The 12PN flux coefficients beyond the printed 4.5PN terms are supplied only through an online repository; an independent recomputation of those coefficients and a direct comparison of the resulting phase to numerical black-hole perturbation data would test whether the convergence ranking persists."],"forward_implications":["For early-inspiral EMRIs like System1, 8PN TaylorT1 templates keep the two-year dephasing below the O(0.1) rad threshold, so they can serve as accurate and inexpensive templates for matched filtering.","Fully analytic TaylorT2 and TaylorF2 templates need roughly 9PN to reach the same accuracy, but their closed forms make them convenient for template-bank construction and for adding beyond-GR corrections.","For late-inspiral systems near the last stable orbit, all current 12PN templates fail the 0.1 rad / 10^{-3} mismatch target over two years, implying at least 14PN or resummation is needed.","The closeness of TaylorT1a/T4a to TaylorT1/T4 indicates the PN series for the inclination Y(x) converges well in the tested range.","Convergence is slower for larger black hole spin q and larger initial inclination YI, so worst-case template performance should be assessed at q ~ 0.9 and YI ~ 0.9."],"supporting_citations":[{"why":"Numerical flux data used to check consistency of the 12PN energy and angular-momentum loss rates.","marker":"[14]"},{"why":"More recent numerical self-force results against which the 12PN formulas are checked.","marker":"[15]"},{"why":"Supplies the adiabatic PN evolution formulation and the high-order flux machinery this paper extends.","marker":"[19]"},{"why":"Provides the generic-bound-orbit PN secular evolution formulas whose e=0 limit the 12PN results must match.","marker":"[20]"},{"why":"Equatorial high-PN flux results used to check the Y=1 limit and as a source of resummation techniques.","marker":"[22]"},{"why":"Online repository that holds the 12PN coefficients beyond the appended 4.5PN terms; the paper's higher-order results depend on it.","marker":"[31]"},{"why":"Defines the TaylorT1/T2/T3 time-domain approximant families used here.","marker":"[36]"},{"why":"Defines the TaylorF1/F2 frequency-domain approximants and the stationary-phase waveform form.","marker":"[37]"},{"why":"Derives 22PN circular-orbit Schwarzschild templates; provides the earlier benchmark showing TaylorT1 convergence and the LSO pole argument.","marker":"[38]"}],"fun_headline_variants":["12PN phase templates pick TaylorT1 for EMRIs","TaylorT1 leads 12PN EMRI phase templates","New 12PN templates: TaylorT1 wins for EMRI phase","Best EMRI phase template: TaylorT1 at 12PN","12PN order ranks TaylorT1 top for spherical EMRIs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the 12PN flux coefficients above the printed 4.5PN order, which are taken from an online repository rather than derived in the paper, are correct, and that the adiabatic first-order dissipative self-force approximation describes the phase.","fun_headline_variants_meta":{"raw":{"variants":["12PN phase templates pick TaylorT1 for EMRIs","TaylorT1 leads 12PN EMRI phase templates","New 12PN templates: TaylorT1 wins for EMRI phase","Best EMRI phase template: TaylorT1 at 12PN","12PN order ranks TaylorT1 top for spherical EMRIs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000154,"raw_usage":{"total_tokens":1222,"prompt_tokens":965,"completion_tokens":257,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":182}},"tokens_in":581,"tokens_out":257,"duration_ms":3167,"temperature":1.0,"reasoning_tokens":182,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:58:23.031690+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the 12PN flux coefficients for k=10 through 24 using an independent frequency-domain black-hole perturbation code and use them to evolve a two-year TaylorT1 inspiral for q=0.9, YI=0.9 in System1; if the resulting dephasing relative to a fully numerical self-force trajectory exceeds 0.1 rad, or if any coefficient disagrees with the online values, the paper's convergence claim would be overturned.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Online repository that holds the 12PN coefficients beyond the appended 4.5PN terms; the paper's higher-order results depend on it."},{"cited_title":"Comparison of post-Newtonian templates for extreme mass ratio inspirals","cited_arxiv_id":"1304.5675","evidence_quote":"Derives 22PN circular-orbit Schwarzschild templates; provides the earlier benchmark showing TaylorT1 convergence and the LSO pole argument."}],"review_version":1}