{"id":"cea7fadb-d85a-42f9-a566-bcd280236490","arxiv_id":"2603.24661","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A post-Newtonian boosted center-of-mass charge template makes numerical-relativity frame-fixing up to ~25x more robust to the fitting-window choice.","lead":"This paper improves how numerical-relativity gravitational waveforms are moved into a standard center-of-mass frame: it replaces a linear fit with a post-Newtonian formula that also follows the physical wobble of the center of mass. The new fit is much less sensitive to the chosen time window and is now available in the scri package.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Variance-ratio evidence is internal consistency, not accuracy; nuisance parameters α1, α2 can absorb model error so low window-variance does not prove an unbiased CoM frame.","rationale":"The reader's weakest assumption concerns the correctness of the Eq. (26) template, especially the phase mapping and junk/higher-PN contamination. My concern overlaps but is more specific: even if the template is imperfect, the nuisance parameters α1 and α2 provide a mechanism to absorb those imperfections, decoupling β and Δ from the model error and artificially lowering their window-to-window variance. The paper's central evidence is the variance ratio in Table II, which measures only internal consistency of the fitted parameters, not whether those parameters correspond to the true CoM frame. One can imagine a fit that is very stable across windows but wrong by a constant offset, or a fit whose variance is low because α parameters bend the template to accommodate junk radiation. The start-fixed results in Table II, where the variance improvement nearly vanishes, are consistent with this worry: when the contamination is too strong to be absorbed, the apparent robustness disappears. The paper does not provide an independent check of accuracy, such as injecting a known boost and translation into a synthetic waveform and recovering them. Therefore the central claim, as stated about robustness, is supported but should be treated as conditional on such a validation. This does not change the reader's CONDITIONAL verdict; it strengthens the specific condition that should be met before production use.","tokens_in":16032,"tokens_out":8749,"duration_ms":104775,"concrete_test":"Construct a synthetic benchmark with a known frame: evolve a quasicircular, nonspinning PN trajectory, generate G from Eq. (20) plus a controlled additional 2.5PN n-direction term and/or an injected junk-radiation transient, then apply a known boost β_true and translation Δ_true using the small-boost transformation of G consistent with Eq. (23). Run the same least-squares fit of Eq. (26) over the three window families (start-, center-, end-fixed). Compare recovered (β, Δ) to truth as a function of window size: plot bias and RMSE, not just variance. If parameters converge to β_true, Δ_true for windows ≥1500M even with contaminations, the low variance indicates genuine robustness; if variance is low but bias is large or window-dependent, the central claim fails because robustness is achieved by overfitting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that Eq. (26) makes fitted β and Δ less sensitive to window choice, quantified by Table II variance ratios. But 'variance across windows' is a self-consistency test, not an accuracy test. Eq. (26) is already flexible: the rescaling coefficients α1, α2 can absorb amplitude and phase errors in the leading-PN oscillatory term, and over finite windows α1, α2 are partially degenerate with β and Δ. If the true G contains junk radiation or higher-PN terms not captured by the leading-PN template, the fit can trade those components against α1, α2 and produce β, Δ that are stable across windows yet biased relative to the true frame parameters. The paper's own Fig. 6 shows several simulations with (α1, α2) far from the expected (1,0), and Table II's start-fixed row (ratios 1.1–1.7) is exactly the regime where the template cannot absorb early-time contamination. Thus the largest reported improvements (center-fixed medians ~25 and ~20) may reflect the ability of nuisance parameters to absorb model misspecification, rather than a more accurate CoM frame. Without a ground-truth test, robustness of the fit does not establish correctness of the frame.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an improved method for fixing the center-of-mass (CoM) part of the BMS frame of numerical relativity waveforms. The authors compute a leading-post-Newtonian expression for the boosted CoM charge, Eq. (26), which includes the physical out-spiraling oscillation of the CoM, and fit this template to the numerically computed CoM charge from CCE waveforms. They compare the new fit to the previous linear fit by varying the fitting-window duration and placement for 20 nonprecessing quasicircular SXS simulations. They report median variance ratios up to ~25 for βx and ~20 for Δx for center-fixed windows, and implement the method in the scri package. The central claim is that the PN-based fit yields boost and translation parameters that are substantially less sensitive to the choice of fitting window.","tokens_in":16328,"tokens_out":6581,"duration_ms":61785,"significance":"If the improvement is real and the fitted parameters are unbiased, the method reduces the user-dependence of BMS frame-fixing for CCE waveforms, which is important for waveform modeling and NR/analytic comparisons. The paper draws on external PN results, uses a reasonable set of simulations, and makes its code available through scri. However, the evidence presented is exclusively about internal consistency across windows; the paper does not demonstrate that the new fit produces the correct CoM frame. The nuisance parameters α1 and α2 can absorb template misspecification, so the robustness gain may be partially an artifact. A validation test against an independent known boost/translation is needed before the method is adopted as a definitive frame-fixing tool.","major_comments":[{"comment":"The headline improvement is measured as a ratio of variances of fitted parameters over different windows. This is a self-consistency metric; it does not establish that the fitted β and Δ equal the true CoM-frame parameters. Equation (26) contains two nuisance parameters α1 and α2 that can absorb amplitude and directional errors in the leading-PN template. A misspecified template can produce parameters that are stable across windows but biased. The paper's own Fig. 6 shows several simulations with (α1, α2) far from (1,0), and the text says the behavior of these outliers is not addressed. To support the frame-fixing claim, please add a recovery test (e.g., apply a known boost/translation to a waveform and check the fitted parameters) or compare with an independent CoM estimate. Without such validation, the central claim is about precision, not accuracy.","section":"Section VI, Table II and Fig. 6"},{"comment":"The error-order statements are inconsistent: Eq. (25) says O(x, β^2) while Eq. (26) says O(x^3, β^2). Since G ~ x^{5/2} and J ~ x^{1/2}, an O(x) remainder in Eq. (25) is not the natural PN order of the neglected terms; the boost terms omitted from Eq. (23) are of order β x^{1/2}. Please clarify the PN counting: what exactly is being neglected in each equation, and why does the remainder improve from O(x) to O(x^3) between the two equations?","section":"Section V, Eqs. (25)–(26)"},{"comment":"The coefficient 1142/105 in the total CoM flux is a key input to the template, but its derivation is only summarized in one sentence. If this coefficient is not already present in the cited Compère et al. expression, the authors should show the intermediate steps or give a direct reference. Currently a reader cannot verify the central analytical result on which the fitting function rests.","section":"Section V, Eq. (19)"}],"minor_comments":[{"comment":"Report the spread (e.g., quartiles or confidence intervals) of the variance ratios across the 20 simulations, not just the median.","section":"Table II"},{"comment":"The caption states the simulation, but the vertical axis labels are in different units for β and Δ; consider adding a legend or an explicit note about the different scales.","section":"Fig. 5"},{"comment":"Typo: 'Conversly' should be 'Conversely'.","section":"Section VI"},{"comment":"State the branch of arg and the convention used to unwrap ψ to avoid phase jumps, since the fitting is sensitive to phase continuity.","section":"Section V, Eq. (28)"}],"recommendation":"major_revision","confidential_remarks":"The paper is promising and the robustness data are useful, but the missing accuracy validation is the main obstacle. I would encourage the editor to ask for a recovery or cross-validation test. The PN-order inconsistency in Eqs. (25)–(26) also needs correction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a solid, well-scoped incremental improvement to BMS frame-fixing for CCE waveforms, and the central claim—that the PN-boosted fit reduces window sensitivity—holds up. The new bits are the boosted fitting function (Eq. 26) with the two nuisance parameters, the derivation of the leading-PN CoM charge from the Compère balance law, and a 20-simulation variance-ratio analysis showing the previous linear fit is far more window-sensitive. The paper also ships the method in scri, which is useful.\n\nThe paper does what it claims: it demonstrates robustness of the fit parameters to window choice, with median variance-ratio ~25 for beta_x and ~20 for Delta_x in the center-fixed case. The analytical function visibly fits the out-spiral in Fig. 2. The authors are honest about the limits: they exclude q close to 1 because of the mass-ratio prefactor, note junk radiation limits the start-fixed case, and flag the unexplained scatter in alpha1, alpha2.\n\nSoft spots, in order of real importance. First, the variance ratio is an internal-consistency metric, not an accuracy metric. As the stress-test note says, alpha1 and alpha2 are fit to the same data and can absorb model error; stable across windows doesn't prove the frame is unbiased. The paper doesn't overclaim—it says robustness—but for production catalogs you'd want some check against an external reference (e.g., PN predictions for the boost or cross-comparison with horizon-based CoM). Second, the derivations are abbreviated: the step to 1142/105 in Eq. (19) is summarized as 'we used the equations of motion' with no intermediate algebra, and the PN-order counting in Eq. (25) vs Eq. (26) looks inconsistent (O(x,beta^2) then O(x^3,beta^2)) without explanation. These are fixable with a derivation appendix. Third, Table II reports medians without spreads or error bars; the claim of '~25' for the boost vector is really only beta_x, with beta_y at 17.6 and Delta_y 11.1—still good, but the abstract overstates the uniformity. Finally, the CCE data isn't public yet, so the 20-simulation analysis isn't independently reproducible today.\n\nNone of this undercuts the main result. This deserves a serious referee.","headline":"Solid, well-scoped improvement to BMS CoM frame-fixing: the PN-boosted fit does reduce window sensitivity, though the variance-ratio metric measures internal consistency, not unbiasedness.","tokens_in":16900,"tokens_out":2649,"would_cite":true,"duration_ms":27519,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Fitting the center-of-mass charge to a boosted post-Newtonian template makes numerical-relativity waveform frame fixing much less sensitive to the analyst's choice of fitting window, with median variance improvements up to about a factor of","keywords":["numerical relativity","BMS frame fixing","center-of-mass charge","post-Newtonian approximation","gravitational waveforms","binary black holes","Cauchy-characteristic evolution","waveform gauge"],"falsifier":"Take a long, low-junk simulation and compute the boost and translation from the PN fit using windows anchored at early, middle, and late times; also compute the CoM frame independently from apparent-horizon trajectories. If the PN-fit parameters disagree with the horizon-based frame by more than the linear fit's disagreement, or if the variance ratios are near unity once junk-free early data are excluded, the central robustness claim fails.","tokens_in":15891,"feed_emoji":"🌌","tokens_out":4786,"duration_ms":44330,"temperature":0.7,"pith_summary":"This paper tries to make BMS (Bondi–Metzner–Sachs) frame fixing of numerical-relativity gravitational waveforms less dependent on the analyst's choice of fitting window. In previous work, the center-of-mass (CoM) charge was fit with a straight line, even though it physically spirals outward during the inspiral. The authors derive a leading-order post-Newtonian expression for the boosted CoM charge and use it as the fitting template, with nuisance parameters to absorb amplitude and direction mismatches. They report that the resulting boost and translation parameters are substantially less sensitive to window duration and placement, with median variance reductions of up to about a factor of 25 for the boost and 20 for the translation. If correct, this makes BMS-frame-fixed waveforms more reproducible and easier to compare with analytic models.","feed_headline":"Frame fixing gets 25x more stable with a post-Newtonian boost model","feed_subtitle":"A post-Newtonian template for the center-of-mass charge cuts sensitivity to fitting-window choice by up to 25x.","key_machinery":"The central object is the boosted CoM charge fitting function, Eq. (26): G' = (1/P^0)[(alpha1 lambda-hat + alpha2 n-hat)|G| - beta x J - u P^0 beta] + Delta. It packages the leading-PN CoM charge |G| (proportional to x^{5/2} sqrt(1-4 nu) nu^2), the spin-weighted orbital direction vectors, a small-boost transformation involving the angular momentum, and nuisance parameters that absorb higher-PN and metadata errors. Its role is to let the fit separate the physical out-spiral from the spurious boost and translation, making the extracted beta and Delta robust to the fitting-window choice.","core_discovery":"The authors claim that the numerical CoM charge of a quasicircular, nonprecessing binary, once boosted and translated by the unknown frame parameters, is well fit over the inspiral by the leading-PN analytical form of Eq. (26). That template combines the PN out-spiral of the CoM charge, evaluated with orbital phase taken from the (2,1) strain mode, with a small-boost transformation and nuisance parameters alpha1, alpha2, plus a constant translation Delta. Fitting this template instead of a line yields boost and translation parameters that vary far less as the fitting window is moved or resized, and the method is incorporated into the public frame-fixing code so Cauchy-characteristic-evolutio","pith_inferences":["If the PN template's phase mapping from the (2,1) mode is slightly off, the apparent gain could partly reflect overfitting rather than a true frame; a direct test would compare with an independent CoM determination.","The same template idea should extend to eccentric and precessing binaries once the corresponding PN CoM-charge expressions are derived; the nuisance-parameter structure may need to grow to absorb in-plane higher-PN effects.","The improvement largely vanishes when the window is anchored at the start of the inspiral, suggesting junk radiation, not the linear-drift baseline, is the main contaminant; treating junk radiation explicitly could yield further gains.","Higher-PN corrections to the CoM charge might shrink the scatter of the nuisance parameters around (1,0), turning the fit into a diagnostic for PN-numerical-relativity agreement."],"forward_implications":["Frame-fixed Cauchy-characteristic-evolution waveforms from numerical relativity will depend less on which window an analyst chooses, improving reproducibility.","Users can safely use shorter windows (at least about 1500M) and center them in the inspiral, reducing contamination from junk radiation and merger.","Waveform models calibrated or compared against numerical-relativity waveforms will see smaller gauge-induced mode mixing, tightening comparisons.","The z-component of the CoM charge still requires a linear fit, so out-of-plane frame fixing remains as before.","The method is now incorporated into the public Python package used for asymptotic waveform analysis, making it directly usable by the community."],"fun_headline_variants":["PN boost model stabilizes waveform frame by 25x","25x more robust frame fits with PN center-of-mass charge","Post-Newtonian trick boosts numerical relativity frame stability","New template cuts frame-fit variance by factor of 25","PN charge model tightens gravitational wave frame fixing"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The template assumes the numerical CoM charge, after rescaling by nuisance amplitudes, is accurately captured by the leading-PN out-spiral with phase taken from the (2,1) mode; junk radiation or higher-PN corrections can violate this and bias the fit.","fun_headline_variants_meta":{"raw":{"variants":["PN boost model stabilizes waveform frame by 25x","25x more robust frame fits with PN center-of-mass charge","Post-Newtonian trick boosts numerical relativity frame stability","New template cuts frame-fit variance by factor of 25","PN charge model tightens gravitational wave frame fixing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000127,"raw_usage":{"total_tokens":968,"prompt_tokens":779,"completion_tokens":189,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":523,"completion_tokens_details":{"reasoning_tokens":110}},"tokens_in":523,"tokens_out":189,"duration_ms":3400,"temperature":1.0,"reasoning_tokens":110,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T05:40:03.869550+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a long, low-junk simulation and compute the boost and translation from the PN fit using windows anchored at early, middle, and late times; also compute the CoM frame independently from apparent-horizon trajectories. If the PN-fit parameters disagree with the horizon-based frame by more than the linear fit's disagreement, or if the variance ratios are near unity once junk-free early data are excluded, the central robustness claim fails.","supporting_citations":[],"review_version":2}