{"id":"d462f607-0ce8-4c1d-9cc2-2763deef51a7","arxiv_id":"2608.02747","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Using 3D supernova simulations, the authors predict that neutrino memory creates a gravitational wave background with Omega_GW around 1e-16 at 0.1 Hz, within reach of future space-based detectors.","lead":"Scientists modeled the gravitational wave hum from all the supernovae in the universe and found a low-frequency 'memory' component from the neutrinos these explosions emit. This signal peaks near 0.1 Hz, right in the band of future space-based detectors, so it could hide or mix with gravitational waves from the early universe.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The predicted 0.1 Hz peak and Omega_GW ~ 1e-16 depend on an extrapolation of the fitted memory waveform beyond the 4.5 s simulation window, so the comparison to inflationary backgrounds is not yet secured.","rationale":"I read the paper as a first quantitative estimate of the supernova neutrino-memory stochastic gravitational wave background, with limitations that are honestly stated. The central claim is plausible, and the orientation and star-formation-rate robustness checks add some support. The load-bearing weakness is the unvalidated extrapolation from 4.5 s to ~10 s post-bounce. I agree with the reader's weakest assumption and sharpen it: the issue is not merely missing data, but that the fitted basis functions are allowed to have long tails and that exactly the timescales missing (5-10 s) can contribute coherently at 0.1 Hz. This does not require rejecting the paper; it requires releasing the fitted parameters and running the truncation or longer-simulation test, which is precisely a conditional verdict. Hence no verdict change is needed.","tokens_in":21808,"tokens_out":6377,"duration_ms":67041,"concrete_test":"Recompute the population-averaged Omega_GW with each fit truncated at its simulation end (holding h constant after t_end) rather than using the full erf extrapolation; compare the 0.1 Hz peak amplitude and f_peak. If the truncated peak is within a factor of 2 of the reported ~1e-16, the extrapolation is not load-bearing; if it is lower by a larger factor, the headline signal is an artifact of the assumed late-time ansatz. As a second check, repeat the analysis using a longer 3D core-collapse supernova simulation (e.g., a model evolved to at least 7 s post-bounce, such as Bollig et al. 2021) and test whether the same fitted functional form reproduces the memory strain at 5-7 s.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the fitted exponential-plus-Gaussian memory model (Eqs. 8-13) correctly describes neutrino memory over the first ~10 s after bounce. The BWV simulations used for the fits extend only to ~4.5 s (Table I and Sec. III A), and the paper itself concedes in Sec. V that no 3D core-collapse supernova simulations of adequate duration exist. This matters because the 0.1 Hz band is sensitive to anisotropy variations on ~1 s timescales at any time during the burst: a Gaussian anisotropy centered at t ~ 7 s with sigma ~ 1 s contributes to the same frequency band with a phase, not an amplitude, suppression (Eq. 11). The fit therefore cannot constrain late-time cooling-phase anisotropies at 5-10 s, exactly where Eq. (8) is an assumed exponential decay. Additionally, the fit allows zeta_j values down to 0.01 Hz (Sec. III B), corresponding to sigma ~ 70 s, so individual fitted Gaussians can extrapolate the asymptotic memory amplitude far beyond the simulated window even when their centers lie inside it. BIC model selection tests in-window goodness of fit, not the validity of the functional form after t_end. If late-time anisotropies add or cancel memory, the 0.1 Hz peak height, and thus the SNR (Table II) and the claimed proximity to the inflation bound, shift.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript estimates the stochastic gravitational-wave background (SGWB) from core-collapse supernovae, emphasizing the neutrino-memory component. The authors fit a phenomenological model—an exponential neutrino luminosity decay plus a sum of up to 40 Gaussians for the emission anisotropy (Eqs. 8–13)—to the 3D BWV simulation suite [56], which extends to ~4.5 s post-bounce, and use the fits to compute single-progenitor and population-averaged spectra via Eq. (2). They find that the memory component peaks at f ~ 0.1 Hz with Omega_GW ~ 1e-16, potentially detectable by DECIGO and BBO with SNR 1.4–10.3 over one year, and comparable to the maximum allowed slow-roll inflationary background. The matter component peaks at ~1 kHz and is far below detectability. The paper concludes that the memory background may constitute a 'neutrino fog' for cosmological SGWB searches.","tokens_in":22100,"tokens_out":7773,"duration_ms":70166,"significance":"If the peak prediction is robust, this is an important result: it identifies a guaranteed astrophysical foreground in the decihertz band, quantifies its contribution with a state-of-the-art 3D simulation suite, and gives a concrete target for DECIGO/BBO science. The manuscript is transparent about its main limitation—no existing 3D simulations cover the full memory evolution—and it tests sensitivity to observer orientation and to the star-formation-rate model. Because the prediction is obtained by extrapolating fitted waveforms beyond the simulation window, the claimed proximity to the inflationary bound must be supported by additional robustness tests before the result can be considered secure.","major_comments":[{"comment":"The central prediction rests on an extrapolation of the fitted memory waveform beyond the simulation duration. The BWV simulations used for the fits extend to at most 4.5 s post-bounce (Table I), and Sec. V states that no 3D simulation of adequate duration exists to capture the full memory evolution. Eq. (8) assumes an exponential decay valid for 'the first ~10 s or so', but the fit itself can only constrain the model within the simulation window. Because the 0.1 Hz band is sensitive to anisotropy variations on ~1 s timescales at any time during the burst (see Eq. 11), an unmodeled late-time anisotropy at t ~ 5–10 s would contribute to the same band. The BIC-based selection of N_best tests in-window goodness of fit, not the validity of the exponential-plus-Gaussian functional form after t_end. I request a quantitative robustness test: inject synthetic late-time anisotropy components (e.g., Gaussians with centers at 5–10 s and amplitudes comparable to fitted in-window components) and recompute the population-averaged Omega_GW and the SNR values in Table II, reporting the resulting spread in the 0.1 Hz peak amplitude.","section":"Sec. III A, Table I; Eq. (8); Sec. V"},{"comment":"The domain constraint 0.01 <= zeta_j <= 100 Hz permits Gaussians with sigma ~ 70 s, whose support extends far beyond the simulation window even when their centers lie inside it. Consequently, fitted components can extrapolate asymptotic memory amplitude well past t_end, and the predicted decihertz peak may be partly an artifact of this allowed domain rather than a feature resolved in the 4.5 s of simulation. The paper should report the distribution of best-fit zeta_j (or sigma_j) values across the 20 models and both polarizations, and it should recompute the population-averaged spectrum with an upper bound on sigma tied to the simulation duration (e.g., sigma <= ~1 s or a corresponding cap on zeta_j) to show that the 0.1 Hz peak persists.","section":"Sec. III B"},{"comment":"The abstract and discussion claim that the memory background may have energy density comparable to the maximum slow-roll inflationary background (Omega_GW ~ 3e-16). Given that the peak amplitude is set by the extrapolated late-time behavior (see major comment 1), this comparison is not yet secured. The manuscript should provide either a lower bound on the memory peak obtainable from the in-window data alone (e.g., by truncating the fitted waveforms at t_end) or an explicit uncertainty band on the curve in Fig. 6 that includes the late-time extrapolation uncertainty, before concluding that the SN-SGWB can affect inflationary searches.","section":"Sec. IV B, Fig. 6; Sec. V"},{"comment":"The paper says the estimate for the neutrino memory component 'may be considered conservative'. This directional claim is not supported by the analysis: unmodeled late-time anisotropy can either increase or decrease the net memory strain, and the absence of rotation (which the paper expects to enhance the signal) also introduces a one-sided bias. The word 'conservative' should be removed or substantiated with a one-sided robustness test.","section":"Sec. V"}],"minor_comments":[{"comment":"The organization paragraph states 'In section I, we introduce the numerical simulations adopted here', but the simulations and fits are presented in Section III; please correct the cross-reference.","section":"Sec. I (Introduction), Sec. III"},{"comment":"The phrase 'space-born' should be 'space-borne' (e.g., in the abstract and the introduction).","section":"Abstract and throughout"},{"comment":"There is a typographical artifact '˙These' before 'These competing signals'; please fix the punctuation.","section":"Sec. IV C"},{"comment":"Equation (14) appears to contain a typographical error in the exponential factor of the high-frequency term; it should likely be e^{-f_s/b} rather than the garbled 'e^{-fe b}', and the bracket structure should be checked.","section":"Sec. II C, Eq. (14)"},{"comment":"The term 'wl4GNZ model' is introduced without a definition or reference; please clarify what this model is and how Eq. (8)–(9) relate to Ref. [51].","section":"Sec. II B"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope for the journal and addresses a timely topic. The main concern is the extrapolation beyond the simulation window, which is load-bearing for the headline claims; this can be addressed with robustness tests. The authors' transparency about the simulation-duration limitation is commendable and makes the revision path clear."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a legitimate first population-averaged estimate of the neutrino-memory SGWB from modern 3D supernova simulations, with a clear decihertz peak around Omega_GW ~1e-16. The paper is honest about its main weakness — no simulation covers the full memory evolution — but that weakness is load-bearing: the peak amplitude depends on an extrapolation beyond the 4.5 s simulation window, so the comparison to inflationary backgrounds is a reasonable estimate, not a secure prediction.\n\nWhat's actually new: prior work (Buonanno et al. 2005) identified the memory SGWB in principle, and later papers modeled the matter component. This is the first time the memory component is computed with a population of twenty 3D multi-second models, with mass weighting, orientation checks, and detector sensitivity comparisons. The fits capture the large-scale waveform features that set the sub-Hz peak, and the authors check orientation averaging and SFR variations carefully. The SNR numbers (1.4 for DECIGO, 10.3 for BBO) are not oversold. The writing is clear, and the limitations are stated directly, including the admission that no current 3D simulation covers the full memory evolution.\n\nThe soft spot is real, though. The BWV simulations stop at ~4.5 s, and the 0.1 Hz band is sensitive to anisotropy variations on ~1 s timescales anywhere in the burst. The phenomenological model allows Gaussians with sigma up to ~70 s, so individual fitted components can extrapolate the asymptotic memory far past the simulated window. BIC selects fits that look good inside the window; it does not validate the functional form after t_end. If cooling-phase anisotropies behave differently at 5–10 s — adding or canceling memory — the peak height shifts, and with it the SNR and the comparison to inflation. The paper acknowledges this in spirit, but the central claim leans on it. Also, there are no error bars on the peak, and everything derives from one simulation suite.\n\nBottom line: the qualitative result — a decihertz memory foreground near 1e-16 — is physically well motivated and likely robust. The quantitative comparison to inflationary backgrounds is not yet secured. This deserves a serious referee. I would want the authors to release the fitted parameters, propagate uncertainties, and test with a longer-duration or second simulation suite before the inflation comparison is treated as established. For anyone modeling future decihertz detectors, this is a useful paper to know about, and I would cite it with that caveat. Send it to peer review.","headline":"A useful first population-averaged estimate of the neutrino-memory SGWB, with a decihertz peak near Omega_GW ~1e-16 — plausible, but the claimed proximity to inflationary backgrounds depends on an extrapolation past the simulation duration.","tokens_in":22674,"tokens_out":2338,"would_cite":true,"duration_ms":23546,"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":"Core-collapse supernovae are predicted to produce a stochastic gravitational-wave background from neutrino memory that peaks near 0.1 Hz at a level observable by DECIGO and BBO within a year.","keywords":["stochastic gravitational wave background","supernova neutrinos","gravitational wave memory","core-collapse supernovae","decihertz gravitational waves","DECIGO","BBO","neutrino fog"],"falsifier":"Search one year of DECIGO or BBO auto-correlation data near $f \\sim 0.1$ Hz after subtracting the compact-binary foreground: a null result with sensitivity below $\\Omega_{\\rm GW}\\sim 10^{-16}$ would rule out the fiducial supernova-memory background. Alternatively, a 3D core-collapse simulation run beyond 10 seconds post-bounce whose neutrino memory spectrum differs markedly from the exponential-plus-Gaussians fit would break the extrapolation that sets the decihertz peak.","tokens_in":21556,"feed_emoji":"🌊","tokens_out":11158,"duration_ms":92024,"temperature":0.7,"pith_summary":"This paper argues that the cumulative, unresolved population of core-collapse supernovae generates a stochastic gravitational-wave background whose most promising piece is the gravitational-wave memory from anisotropic neutrino emission, not the matter motion inside the exploding star. That memory component peaks at $f \\sim 0.1$ Hz with $\\Omega_{\\rm GW}\\sim 10^{-16}$, putting it within reach of planned space-based decihertz detectors: the paper estimates signal-to-noise ratios of 1.4 for DECIGO, 3.1 for an improved DECIGO, and 10.3 for BBO in one year of data. At that amplitude the supernova background is comparable to the maximum slow-roll inflationary background allowed by current CMB bounds and to some phase-transition and cosmic-string relic spectra, so searches for those cosmological signals would have to account for or subtract it. The matter component, peaking near $10^3$ Hz at $\\Omega_{\\rm GW}\\sim 10^{-13}$, is far below the reach of ground-based detectors and is not the main story. The result reframes supernova neutrino emission as a gravitational 'neutrino fog' for future cosmological gravitational-wave searches.","feed_headline":"Neutrino memory may fog the 0.1-Hz gravitational-wave band","feed_subtitle":"This supernova background would rival inflationary signals at 0.1 Hz, so future searches must subtract it.","key_machinery":"The load-bearing object is the gravitational-wave memory strain from neutrinos, $h_i(t,\\Omega) = \\frac{2G}{rc^4}\\int_{-\\infty}^{t-r/c} dt'\\, L_\\nu(t')\\,\\alpha_i(t',\\Omega)$, where $L_\\nu$ is the total neutrino luminosity and $\\alpha_i$ is the dimensionless anisotropy parameter for each polarization. The paper adopts the phenomenological form $L_\\nu(t)=\\beta e^{-\\chi t}$ and $\\alpha_i$ as a sum of Gaussians, which yields the strain as a sum of error functions and the frequency-domain strain as $\\tilde h(f)\\propto \\frac{1}{f}\\exp(-\\pi^2 f^2/\\zeta_j^2)e^{2\\pi i f \\tau_j}$. This simple form carries the argument: the multi-second decay time sets the $\\sim 0.1$ Hz peak, the $1/f$ behavior produces the $\\Omega_{\\rm GW}\\propto f$ low-frequency tail, and the Gaussian widths control the higher-frequency structure; the population integral over redshift and progenitor mass then converts these single-supernova spectra into the stochastic background.","core_discovery":"The central claim is that the stochastic gravitational-wave background from core-collapse supernovae is dominated, in its detectable band, by the neutrino memory effect: the permanent displacement left in the metric by anisotropic neutrino emission during and after the explosion. Fitting a decaying-exponential-plus-Gaussians memory model to a suite of twenty 3D multi-second core-collapse simulations and integrating over the initial mass function and a cosmic star formation history, the paper finds that the population-averaged memory component peaks at $f \\sim 0.1$ Hz with $\\Omega_{\\rm GW}\\sim 10^{-16}$, whereas the matter component peaks near $10^3$ Hz with $\\Omega_{\\rm GW}\\sim 10^{-13}$. The decihertz peak is close to the maximum slow-roll inflationary background allowed by current CMB bounds ($\\Omega_{\\rm GW}\\sim 3\\times 10^{-16}$) and is within reach of DECIGO and BBO (one-year SNR 1.4--10.3), making the supernova memory background a credible 'neutrino fog' that future searches for primordial gravitational waves must confront. The paper also identifies a secondary peak near 1 Hz from short-timescale hydrodynamical variations in neutrino emission and a low-frequency $\\Omega_{\\rm GW}\\propto f$ tail that is common to all models.","pith_inferences":["Editorial inference: if the background is at the predicted level, decihertz observatories will need a two-stage subtraction pipeline, first the compact-binary foreground and then the supernova memory background, before any inflationary claim at $0.1$ Hz can stand; the second stage is only as good as the supernova anisotropy model.","Editorial inference: because the simulations stop after about $4.5$ seconds, the predicted peak rests on the extrapolated late cooling phase; a longer-duration 3D simulation suite that shows the memory still growing substantially after that time would shift the peak frequency down, while earlier saturation would push it up.","Editorial inference: the paper's orientation average uses three fixed observer directions; a full angular average over the simulated neutrino-emission anisotropy could smooth or redistribute the $0.01$--$0.1$ Hz structure, which is exactly the band where the peak sits.","Editorial inference: adding rotation and magnetic fields, which these simulations omit, is the natural next stress test; the authors themselves expect rotation to mildly raise the predicted signal, so the fog could be thicker than the fiducial estimate."],"forward_implications":["If the prediction holds, a positive detection of the memory background would be a first measurement of gravitational-wave memory, a yet-unmeasured prediction of general relativity.","The memory background occupies its own decihertz band ($f\\sim 10^{-2}$--$1$ Hz), cleanly separated from the matter component at $\\sim 10^3$ Hz, so the two supernova contributions can be searched for independently.","At $\\Omega_{\\rm GW}\\sim 10^{-16}$ near $0.1$ Hz, DECIGO and BBO could see the background in about one year (SNR 1.4--10.3), while ground-based Cosmic Explorer and Einstein Telescope cannot see the matter component (SNR $\\sim 10^{-29}$).","Searches for inflationary, phase-transition, or cosmic-string backgrounds at the $10^{-16}$ level must include a supernova-neutrino-memory foreground model; the combined spectrum's minimum in the tens-of-hertz range offers a cleaner window for those cosmological searches."],"supporting_citations":[{"why":"Supplies the twenty 3D multi-second core-collapse supernova simulations whose neutrino-memory and matter spectra are fitted and summed into the stochastic background.","marker":"[56]"},{"why":"Provides the decaying-exponential-plus-Gaussians phenomenological memory model and its frequency-domain strain used for the waveform fits.","marker":"[51]"},{"why":"Establishes the supernova stochastic gravitational-wave background formalism and the decomposition into a memory component and a matter component.","marker":"[28]"},{"why":"Supplies the matter-contribution phenomenological model (with PNS oscillation and hydrodynamic terms) that the paper modifies and adopts.","marker":"[29]"},{"why":"Provides the detector noise curves, PLISC sensitivity curves, and SNR formulas used to judge detectability with DECIGO, BBO, CE, and ET.","marker":"[44]"},{"why":"Supplies the slow-roll inflationary stochastic background spectrum used as the comparison target at Omega_GW around 10^-16.","marker":"[21]"},{"why":"Supplies the cosmic star formation rate parameter values used in the fiducial population integral.","marker":"[70]"},{"why":"Supplies the analytic star formation rate parameterization whose shape is used in the population integral.","marker":"[69]"}],"fun_headline_variants":["Neutrino memory from supernovae may fog the 0.1-Hz gravitational-wave band","Supernova neutrino memory creates a stochastic background that rivals inflation at 0.1 Hz","Neutrino fog: supernova memory could hide primordial gravitational waves at 0.1 Hz","Supernova neutrino memory: a new noise floor for gravitational-wave searches near 0.1 Hz","The neutrino fog: supernova memory peaks at 0.1 Hz, potentially masking cosmic relics"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The prediction assumes that the first few seconds of simulated supernova neutrino emission, extrapolated with a decaying exponential to about ten seconds, faithfully represent how the gravitational-wave memory grows; the paper states that no current simulation runs long enough to check this directly.","fun_headline_variants_meta":{"raw":{"variants":["Neutrino memory from supernovae may fog the 0.1-Hz gravitational-wave band","Supernova neutrino memory creates a stochastic background that rivals inflation at 0.1 Hz","Neutrino fog: supernova memory could hide primordial gravitational waves at 0.1 Hz","Supernova neutrino memory: a new noise floor for gravitational-wave searches near 0.1 Hz","The neutrino fog: supernova memory peaks at 0.1 Hz, potentially masking cosmic relics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00024,"raw_usage":{"total_tokens":1552,"prompt_tokens":1010,"completion_tokens":542,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":422}},"tokens_in":626,"tokens_out":542,"duration_ms":6408,"temperature":1.0,"reasoning_tokens":422,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:09:27.766295+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search one year of DECIGO or BBO auto-correlation data near $f \\sim 0.1$ Hz after subtracting the compact-binary foreground: a null result with sensitivity below $\\Omega_{\\rm GW}\\sim 10^{-16}$ would rule out the fiducial supernova-memory background. Alternatively, a 3D core-collapse simulation run beyond 10 seconds post-bounce whose neutrino memory spectrum differs markedly from the exponential-plus-Gaussians fit would break the extrapolation that sets the decihertz peak.","supporting_citations":[{"cited_title":"The gravitational-wave emission from the explosion of a 15 solar mass star with rotation and magnetic fields","cited_arxiv_id":"2406.09691","evidence_quote":"Supplies the analytic star formation rate parameterization whose shape is used in the population integral."}],"review_version":1}