{"id":"c75e3307-ac61-49f6-9927-4df3f0c53354","arxiv_id":"2608.09295","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"GWMemoryLab simulations show the inspiral Christodoulou memory in binary black holes peaks at a total mass that scales roughly inversely with the detector's low-frequency cutoff.","lead":"This paper presents GWMemoryLab, a numerical tool that computes the inspiral gravitational-wave memory for black hole binaries using standard post-Newtonian formulas. It finds that for each detector low-frequency cutoff there is an optimal total binary mass that maximizes the observable memory, with the optimal mass roughly inversely proportional to the cutoff.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (17) assumes a sharp low-frequency cutoff; real memory searches use a noise-weighted matched filter, so the fitted M_peak ∝ 1/f_low may not predict actual optimal target masses.","rationale":"The paper is internally consistent: the convergence test is clean, the code is public, and the inverse scaling can be checked analytically. From Eqs. (5), (10), and (15), the energy radiated between f0 and fISCO is E ∝ η M^{5/3}(f_I^{2/3} - f0^{2/3}); maximizing over M gives M_peak ∝ f0^{-1} with a coefficient close to the fitted 2006.5 (an analytic estimate gives roughly 2043). So the numerical fit is not a fragile empirical accident. The reader's stated weakest assumption, the scalar proxy in Eq. (13), is only partially concerning: a constant angular prefactor in the full tensorial memory would not change the location of the mass peak. The genuinely load-bearing step is the identification of \"observable\" with a hard frequency cutoff rather than with a noise-weighted detector response. That is what would have to be true for Eq. (17) to guide actual searches, and it is not demonstrated in the manuscript. The proposed noise-PSD matched-filter check would settle whether the concern lands. The verdict remains CONDITIONAL: the internal scaling is credible, but its advertised predictive use is conditional on a realistic detector-response calculation.","tokens_in":9790,"tokens_out":13016,"duration_ms":147755,"concrete_test":"Take GWMemoryLab's h_mem(t; M) for equal-mass binaries at D = 400 Mpc and M = 20-200 M_sun. For a chosen detector noise PSD (e.g., aLIGO O4 or ET-D), compute rho^2(M) = ∫_{f_min}^{f_max} |\\tilde h_mem(f; M)|^2 / S_n(f) df, or inject the memory template into simulated noise and perform a matched filter. Locate argmax_M rho. Repeat for several detectors and low-frequency operating points. If the argmax is not within about 20% of 2006.5 / f_low, Eq. (17) is not transferable to real detector-target selection; if it is, the sharp-cutoff idealization is benign.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central application of Eq. (17) is as a predictive tool for selecting masses in future memory searches. That application requires \"observable memory\" to be the quantity a detector actually measures. In the paper, the observable memory is defined as the energy radiated between a sharp low-frequency cutoff f0 and fISCO (§IV.I, Fig. 11). A real detector has a continuous noise spectral density S_n(f); the detectability of a memory signal h_mem(t) is controlled by the matched-filter statistic rho^2 ∝ ∫ |\\tilde h_mem(f)|^2 / S_n(f) df. The memory accumulates as a slowly rising ramp, whose Fourier power is concentrated at low frequencies, so the shape of S_n(f), not just its lower edge, weights different masses differently. Therefore the maximum of rho(M) need not lie at M = 2006.5 / f_low. The scalar proxy in Eq. (13) is not the main bottleneck: at leading order h_mem is linear in E_GW, so the angular prefactor of the full tensor memory would rescale the amplitude but not shift the mass optimum. The untested assumption is that a sharp frequency cutoff is equivalent to a real detector response.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents GWMemoryLab, a modular Python framework for computing the leading-order post-Newtonian inspiral contribution to the nonlinear (Christodoulou) gravitational-wave memory in non-spinning, quasi-circular binary black holes. The framework evolves the inspiral with the quadrupole formula, computes the radiated energy within a detector band [f0, fISCO], and converts it to a characteristic memory strain via h_mem = 4 G E_GW/(c^4 D). The authors validate the implementation with analytical checks and a convergence study, then study how the memory depends on mass ratio, total mass, and detector low-frequency cutoff. The central result is the empirical scaling M_peak = 2006.5/f_low M_sun (Eq. 17), which states that the total mass maximizing the observable inspiral memory is inversely proportional to the detector cutoff frequency.","tokens_in":10048,"tokens_out":11227,"duration_ms":95743,"significance":"If the scaling relation holds, it provides a convenient rule of thumb for selecting target binary masses in searches for inspiral memory. The paper's strengths include the public release of GWMemoryLab with a Zenodo archive, a modular architecture that permits component validation, and a convergence table demonstrating good numerical stability for the fiducial configuration. A further strength, not noted in the paper, is that the inverse scaling is not purely empirical: it follows from the leading-order PN energy spectrum dE/df ∝ M_c^{5/3} f^{-1/3} combined with f_ISCO ∝ 1/M, and the analytic coefficient c^3/(10^{3/2} π G M_sun) ≈ 2040 is close to the fitted 2006.5. The paper is therefore a useful contribution, provided the detector-response limitation and the boundary peak issue are addressed.","major_comments":[{"comment":"The f0 = 10 Hz point used to fit Eq. (17) is not an interior maximum: the text states that for 10 Hz the memory \"continues to increase throughout the investigated mass range\" and reaches its maximum at the edge of the grid, 200 M_sun. Using a boundary value as a peak biases the fitted coefficient 2006.5. The authors should either extend the mass range for f0 = 10 Hz until a true peak appears, or restrict the fit to cutoffs with a well-defined interior peak and state the extrapolated nature of the 10 Hz point.","section":"§IV.I, Fig. 11, Eq. (17)"},{"comment":"The quantity maximized in this paper is the scalar memory h_mem computed from the energy radiated in the sharp band [f0, fISCO]. A real detector's sensitivity is governed by the matched-filter SNR, rho^2 ∝ ∫ |\\tilde{h}(f)|^2 / S_n(f) df, with a continuous noise spectral density S_n(f); the memory's low-frequency spectral content is then weighted by S_n(f), not by a sharp cutoff. Since the paper's stated purpose is to guide \"future memory searches,\" Eq. (17) should not be presented as predicting the most detectable mass until the calculation is repeated with a realistic noise curve (e.g., aLIGO design sensitivity), or the claim should be explicitly restricted to the sharp-cutoff proxy.","section":"§IV.I and Conclusions, Eqs. (17)-(19)"},{"comment":"The convergence study does not report the binary parameters used, and the step counts imply an inspiral duration near 0.9 s at dt = 1e-4. The parameter study includes much shorter inspirals: a 200 M_sun binary with f0 = 10 Hz has a duration of about 0.03 s, giving only ~300 steps at the adopted dt = 1e-4. The claimed relative error of ~1e-7 is therefore not demonstrated for the shortest-inspiral configurations, which are exactly the ones that control the f0 = 10 Hz data point in the fit. The authors should add convergence tests for a high-mass, short-duration case or justify why 300 steps suffice there.","section":"Table I"}],"minor_comments":[{"comment":"The sentence \"The leading-order gravitational-wave strain is follows the quadrupole approximation\" contains a grammatical error; it should read \"The leading-order gravitational-wave strain follows the quadrupole approximation.\"","section":"§II.C"},{"comment":"The paper correctly labels Eq. (13) as an angle-averaged characteristic scalar. It would help to state explicitly that the absolute strain values quoted in Section IV (e.g., 4.08e-22) are this characteristic amplitude, not the strain in any particular detector orientation, and that the fitted coefficient in Eq. (17) consequently inherits this scalar-proxy convention.","section":"§II.E, Eq. (13)"},{"comment":"The text mentions an empirical fit with R^2 ≈ 0.99999 for the mass-ratio dependence, but the fit function and parameters are not given. Please include the fit equation or remove the claim.","section":"§IV.F"},{"comment":"No uncertainties are quoted for the coefficient 2006.5 or the exponent -1. Adding standard errors from the fit would allow readers to judge how well constrained the scaling is.","section":"Eq. (17)"},{"comment":"The introduction motivates memory as a detectable signal; citing recent observational searches for gravitational-wave memory by LIGO/Virgo would strengthen the motivation and connect the detector-cutoff discussion to actual search strategies.","section":"Introduction"}],"recommendation":"major_revision","confidential_remarks":"The paper would be considerably strengthened if the authors replaced the purely empirical presentation of Eq. (17) with the one-line analytic derivation from the quadrupole formula; the fitted coefficient is within ~2% of the analytic value c^3/(10^{3/2} π G M_sun), which would make the result more transparent and less vulnerable to the boundary-point concern. The editor may also wish to verify the novelty claim that this scaling has not been previously reported, since it is a direct consequence of standard PN formulas."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a clean, honest numerical study of inspiral Christodoulou memory in leading-order PN. The genuinely new thing is the empirical relation M_peak = 2006.5 / f_low (Eq. 17), which connects detector low-frequency cutoff to the total mass that maximizes observable inspiral memory. The result is credible within the model, and the framework is reproducible: code is on GitHub/Zenodo, convergence is demonstrated in Table I, and the physical interpretation (competition between luminosity and inspiral duration) is sound.\n\nWhat the paper does well: it is transparent about its assumptions, the numerics look stable, and the parameter study is systematic. The claim that this scaling is new for inspiral-only memory seems correct.\n\nThe soft spots are real but manageable. The biggest one is the detector model: Eq. (17) assumes a sharp low-frequency cutoff, so 'observable memory' is just the energy radiated between f0 and ISCO. A real search uses matched filtering against a noise curve, and the spectral shape of S_n(f) weights different masses differently. The stress-test note is right that this is the main bottleneck; the scalar proxy for the tensor memory is less worrying because the angular prefactor rescales amplitude but shouldn't shift the mass optimum. The paper should at least discuss this and maybe redo the fit with a simplified noise-weighted statistic.\n\nSecond, the 10 Hz case is censored: the peak lies at the edge of the explored mass range (200 Msun), not a true measured peak. Including it in the fit without flagging it is a minor quantitative flaw. And the fit has no error bars on M_peak, so the R^2 = 0.999243 is a bit overinterpreted.\n\nThe fitted constant is empirical, not derived, which the authors acknowledge. So 'predictive tool' is better read as 'rule of thumb for the inspiral-only model.'\n\nWho is this for? People planning parameter-space searches for memory, and anyone building similar inspiral-memory codes. It deserves a serious referee: the physics is standard but the quantitative relation is new and the code is reproducible. With a proper treatment of detector response and the censored point, it could be publishable.\n\nI'd send it to peer review.","headline":"A clean, reproducible numerical study whose main result — the M_peak ~ 1/f_low scaling — is credible within the inspiral model but needs a realistic detector-response check before being used as a predictive tool.","tokens_in":10544,"tokens_out":2662,"would_cite":false,"duration_ms":24729,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["83C35","83C25","83-08"],"pacs":["04.30.-w","04.25.Nx","04.80.Nn"],"model":"deepseek-v4-flash","headline":"A scaling law connects the detector's low-frequency cutoff to the binary mass that maximizes observable inspiral Christodoulou memory: $M_{\\rm peak} = 2006.5 / f_{\\rm low}$.","keywords":["gravitational-wave memory","Christodoulou memory effect","post-Newtonian approximation","binary black hole inspiral","detector low-frequency cutoff","gravitational-wave energy flux","empirical mass scaling"],"falsifier":"At $f_{\\rm low} = 10$ Hz the paper's own data show the memory still increasing at the upper edge of the explored mass range ($M = 200\\,M_\\odot$), while Eq. (17) predicts the peak near $200\\,M_\\odot$; extending the mass grid to 300--400 $M_\\odot$ at 10 Hz would show whether the memory actually turns over there or continues to rise, which would falsify the inverse scaling. A companion check is recomputing the same grid with the full spin-weighted memory tensor to see whether the scalar proxy changes the peak location.","tokens_in":9571,"feed_emoji":"📡","tokens_out":10541,"duration_ms":91790,"temperature":0.7,"pith_summary":"The paper builds a modular numerical pipeline, GWMemoryLab, that computes the leading-order post-Newtonian inspiral of non-spinning binary black holes and accumulates the Christodoulou memory strain from the radiated gravitational-wave energy. It asks which binary parameters make the memory most observable through a given detector, and finds that for a fixed detector bandwidth the memory is maximized at an intermediate total mass: heavy binaries radiate more energy but spend less time in the detector band before reaching the innermost stable circular orbit. The paper's central quantitative result is the empirical scaling $M_{\\rm peak} = 2006.5 / f_{\\rm low}$, with $M_{\\rm peak}$ in solar masses and $f_{\\rm low}$ in hertz, which predicts the optimal mass for a given detector low-frequency cutoff. If correct, this gives a simple targeting rule for memory searches and shows that detector bandwidth matters as much as binary properties.","feed_headline":"Optimal memory mass = 2006.5 / detector cutoff","feed_subtitle":"In solar masses with the cutoff in hertz; a targeting rule for gravitational-wave memory searches.","key_machinery":"The load-bearing object is the scalar proxy $h_{\\rm mem} = 4 G E_{\\rm GW}/(c^4 D)$, which converts the accumulated quadrupole energy $E_{\\rm GW}$ into an angle-averaged characteristic memory strain at luminosity distance $D$; the paper treats this as a phenomenological scale isolating the mass-energy dependence of the full tensorial memory. The rest of the machinery is the leading-order post-Newtonian frequency evolution $\\dot f \\propto f^{11/3}$, the quadrupole luminosity $dE/dt \\propto \\eta^2 x^5$, and the Schwarzschild ISCO cutoff $f_{\\rm ISCO} = c^3/(6^{3/2}\\pi G M)$. Their interplay -- more mass means more luminosity but a shorter in-band inspiral -- is what produces the peak in $h_{\\rm mem}$ versus $M$.","core_discovery":"Within the leading-order post-Newtonian approximation for non-spinning, quasi-circular binaries, the accumulated inspiral Christodoulou memory is computed as $h_{\\rm mem} = 4 G E_{\\rm GW}/(c^4 D)$, so the memory tracks the total radiated energy. The paper shows that, for a fixed detector low-frequency cutoff, this observable memory is not monotonic in total mass: increasing mass raises the gravitational-wave luminosity, but it also lowers the ISCO frequency ($f_{\\rm ISCO} \\propto M^{-1}$), shortening the part of the inspiral that falls inside the detector band. The balance produces a distinct optimal mass $M_{\\rm peak}$ for each cutoff, and a fit across cutoffs in the 10--40 Hz range gives $M_{\\rm peak} = 2006.5 / f_{\\rm low}$ in solar masses (with $f_{\\rm low}$ in Hz), with coefficient of determination $R^2 = 0.999243$. The paper presents this inverse scaling as a new quantitative connection between detector bandwidth and the binary systems that yield the largest observable inspiral memory, noting that such a scaling has not been explicitly reported for inspiral-only nonlinear memory studies.","pith_inferences":["An implication the authors leave implicit is that the same inverse scaling can be used to choose target masses for future low-frequency detectors: at $f_{\\rm low} = 5$ Hz it predicts $M_{\\rm peak} \\approx 400\\,M_\\odot$, well outside the range simulated here, so the low-frequency end of the scaling is currently untested.","A testable extension is to replace the scalar proxy with the full angle-resolved memory tensor; if the angular structure changes the energy-to-memory mapping, detectors with non-uniform antenna patterns could see a spread of optimal masses rather than a single value.","Higher-order post-Newtonian corrections, spin, and eccentricity are likely to preserve the inverse functional form but rescale the coefficient 2006.5, so the exact number should be treated as a leading-order estimate."],"forward_implications":["At a given cutoff $f_{\\rm low}$, the observable inspiral memory is largest for total mass about $2006.5/f_{\\rm low}$ solar masses, so lower-cutoff detectors should target heavier black-hole binaries.","Equal-mass binaries produce the largest memory because the symmetric mass ratio is maximal ($\\eta = 1/4$), while highly unequal ratios suppress the radiated energy and memory.","Lowering the cutoff from 40 Hz to 10 Hz increases the accumulated memory by more than an order of magnitude because a much longer early inspiral is included in the observed band.","The non-monotonic mass dependence implies that the heaviest binaries are not the best memory sources; the optimum balances luminosity against in-band inspiral duration."],"supporting_citations":[{"why":"Establishes that gravitational-wave self-interaction produces a permanent nonlinear memory, the effect under study.","marker":"[3]"},{"why":"Derives the Christodoulou memory amplitude from radiated gravitational-wave energy, underpinning Eq. (13).","marker":"[4]"},{"why":"Supplies the hereditary and energy-flux analysis connecting memory to the past history of gravitational-wave emission.","marker":"[5]"},{"why":"Provides the memory waveform and energy relation for inspiralling binaries, used for validation and scaling arguments.","marker":"[6]"},{"why":"Gives nonlinear memory from binary black hole systems as a function of radiated energy, supporting the mass-ratio and total-mass scalings.","marker":"[7]"},{"why":"Supplies the leading-order post-Newtonian evolution equations and energy flux used in the numerical integration.","marker":"[9]"},{"why":"Gives the quadrupole gravitational-wave luminosity formula used to compute the energy flux.","marker":"[10]"},{"why":"Provides the standard post-Newtonian inspiral waveform and ISCO frequency expressions used in the pipeline.","marker":"[11]"}],"fun_headline_variants":["Optimal memory mass scales inversely with detector cutoff","GW memory: peak mass ~ 2000 divided by low cutoff","Inspiral memory: inverse law links mass and detector band","Maximize memory: choose mass by detector's low cutoff","Memory detectability: simple rule for optimal binary mass"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All quantitative memory amplitudes and the fitted coefficient 2006.5 rest on representing the Christodoulou memory by the angle-averaged scalar $4 G E_{\\rm GW}/(c^4 D)$; if the true direction-dependent tensor memory has a different energy-to-strain relation, the optimal masses could shift.","fun_headline_variants_meta":{"raw":{"variants":["Optimal memory mass scales inversely with detector cutoff","GW memory: peak mass ~ 2000 divided by low cutoff","Inspiral memory: inverse law links mass and detector band","Maximize memory: choose mass by detector's low cutoff","Memory detectability: simple rule for optimal binary mass"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000174,"raw_usage":{"total_tokens":1312,"prompt_tokens":1002,"completion_tokens":310,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":230}},"tokens_in":618,"tokens_out":310,"duration_ms":3916,"temperature":1.0,"reasoning_tokens":230,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:56:17.653884+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"At $f_{\\rm low} = 10$ Hz the paper's own data show the memory still increasing at the upper edge of the explored mass range ($M = 200\\,M_\\odot$), while Eq. (17) predicts the peak near $200\\,M_\\odot$; extending the mass grid to 300--400 $M_\\odot$ at 10 Hz would show whether the memory actually turns over there or continues to rise, which would falsify the inverse scaling. A companion check is recomputing the same grid with the full spin-weighted memory tensor to see whether the scalar proxy changes the peak location.","supporting_citations":[{"cited_title":"Christodoulou, Nonlinear Nature of Gravitation and Gravitational-Wave Experiments, Phys","cited_arxiv_id":null,"evidence_quote":"Establishes that gravitational-wave self-interaction produces a permanent nonlinear memory, the effect under study."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Derives the Christodoulou memory amplitude from radiated gravitational-wave energy, underpinning Eq. (13)."},{"cited_title":"Favata, The Gravitational-Wave Memory Effect, Class","cited_arxiv_id":null,"evidence_quote":"Provides the memory waveform and energy relation for inspiralling binaries, used for validation and scaling arguments."},{"cited_title":"Favata, Nonlinear gravitational-wave memory from binary black hole mergers, Astrophys","cited_arxiv_id":null,"evidence_quote":"Gives nonlinear memory from binary black hole systems as a function of radiated energy, supporting the mass-ratio and total-mass scalings."},{"cited_title":"Blanchet, Gravitational Radiation from Post- Newtonian Sources and Inspiralling Compact Binaries, Living Rev","cited_arxiv_id":null,"evidence_quote":"Supplies the leading-order post-Newtonian evolution equations and energy flux used in the numerical integration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the quadrupole gravitational-wave luminosity formula used to compute the energy flux."}],"review_version":1}