{"id":"553266b2-92ff-4e3c-9b60-1fa2f2ee1f4e","arxiv_id":"2505.22736","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Using NANOGrav 15-year and IPTA second data release, this paper sets a 68% confidence lower bound of 10^-19 GeV on the Lorentz-violating scale M_LV in a modified gravity model.","lead":"This study derives how a Lorentz-violating modification of gravity changes the gravitational-wave background spectrum and uses pulsar timing array data to set a lower bound on the energy scale of the modification. The bound, M_LV > 10^-19 GeV, is two orders of magnitude stronger than previous binary merger constraints, showing that nHz gravitational-wave observations can probe fundamental spacetime symmetries.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central bound hinges on an unreported horizon-entry scale a_e and a sign error in Eq. (3.8); the claimed M_LV constraint may be an artifact of this choice.","rationale":"The strongest_claim is exactly the M_LV > 10^-19 GeV constraint, and the reader's weakest_assumption correctly identifies Eq. (3.8), the unspecified a_e, and the sign/damping inconsistency. My independent reading of the manuscript confirms these are load-bearing: the exponential factor in Eq. (3.8) is the only place where M_LV enters the SGWB spectrum, and the likelihood analysis via PTArcade fits the spectrum amplitude A and slope γ while M_LV controls the shape through that exponential. The a_e value is never given, the text says 'a_e refers to the scale factor at horizon entry' but provides no formula or numerical value, and the phrase 'scale-dependent damping' contradicts the positive exponent for a_e < 1. The sign issue is real but secondary because the dominant uncertainty is the magnitude of (a_e^{-2}-1). I do not flag the LIGO/Virgo comparison itself as a concern because the paper explicitly cites [37] for the 10^-21 GeV number, and comparing a PTA lower bound to a binary-merger bound is a legitimate claim even if the models differ. Similarly, I do not treat the hand-set ᾱν = 1 as a fatal flaw by itself; it is a modeling choice, but it is coupled to the a_e issue because ᾱν and a_e both set the effective damping scale. The recommendation is CONDITIONAL rather than REJECT: the framework and the data analysis pipeline are plausible, and the missing pieces (a_e computation, sign justification) are checkable, but the headline number cannot be verified as presented. The reader's conditional verdict is therefore appropriate, and I find no additional independent objection that would flip it to REJECT or ACCEPT without the requested clarification.","tokens_in":7827,"tokens_out":2679,"duration_ms":24254,"concrete_test":"Recompute the SGWB spectrum entering PTArcade with an explicit, stated a_e(f): derive a_e from the radiation-dominated relation k = a_e H(a_e) (or state the assumed reheating history) at f = 10^-9 Hz, then check whether exp[(a_e^{-2}-1)(2πf/(c M_LV))^2] with the reported M_LV > 10^-19 GeV yields a spectrum within the NANOGrav/IPTA 68% credible band. Also re-run the fit with the sign of the exponent flipped (damping only, exponent negative) and report the resulting M_LV bound; if the bound changes by more than an order of magnitude, the headline constraint depends on the sign choice.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central result, M_LV > 10^-19 GeV, is a lower bound on the Lorentz-violating scale from the damping factor exp[(a_e^{-2}-1)(2πf/(c M_LV))^2] in Eq. (3.8). The paper calls this a damping effect, but for a_e < 1 the exponent is positive, so the factor amplifies the spectrum: at f = 10^-9 Hz with a_e ≪ 1, (a_e^{-2}-1) is huge, and the bound becomes exponentially sensitive to the unspecified value of a_e. The text never states how a_e is computed for PTA frequencies, and states 'our result is obtained with f = 10^-9 Hz... the higher the frequency, the higher the obtained bound' — yet Eq. (3.8) has M_LV in the denominator of the exponent, so for fixed data the lower bound on M_LV should be set by the lowest frequency; the quoted frequency dependence is consistent with the exponent, but the numerical value of the many-orders-of-magnitude exponent depends entirely on a_e. There may also be a missing sign: Eq. (2.8) gives H ν̄ ≈ (c1 k^2/a^2)' with the plus sign, and Eq. (3.4) has D = -1/2 ᾱν (k/(a M_LV))^2 evaluated from a_e to a, yielding the a_e^{-2} - 1 form; if a_e were instead defined at the end of inflation and the damping evaluated after horizon entry, the sign convention must be justified. Since the likelihood in PTArcade uses the SGWB amplitude through h_c(f) ∝ exp(D), an incorrect a_e or sign directly rescales the predicted spectrum and therefore the 10^-19 GeV bound. Without the a_e computation, the headline constraint is not reproducible.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a Lorentz-violating modification to the gravitational action by adding extrinsic-curvature derivative terms, derives a modified stochastic gravitational-wave background spectrum, and fits this model to NANOGrav 15-year and IPTA second data release data using PTArcade. It reports a 68% lower bound M_LV > 10^-19 GeV, along with best-fit strain amplitudes and spectral indices for both datasets.","tokens_in":8224,"tokens_out":14614,"duration_ms":161761,"significance":"If the derivation and statistical treatment are correct, the result would strengthen constraints on this particular Lorentz-violating gravity model by about two orders of magnitude over the LIGO/Virgo bound quoted in the paper, and it would demonstrate that pulsar timing arrays can probe the Lorentz-violating scale in this class of theories. A clear strength is that the paper takes a concrete, falsifiable model and confronts it with public PTA likelihoods through a standard tool (PTArcade). However, the central spectral formula is not derived in the manuscript, the sign of the claimed 'damping' effect is inconsistent with the equations as written, and the horizon-entry scale a_e is left unspecified; these issues currently prevent the headline constraint from being reproduced or verified.","major_comments":[{"comment":"The sign of the claimed 'damping' effect is inconsistent with the equations as written. Setting ᾱ_ν=1 makes c1 constant, and Eq. (2.8) then gives Hν̄ = (c1 k^2/a^2)' = -2 H c1 k^2/a^2 < 0. Inserting this into Eq. (3.4) yields D = +1/2 (a_e^{-2} - a^{-2})(k/M_LV)^2 > 0 for a > a_e, so the exponential in Eq. (3.8) amplifies the GW spectrum rather than damping it. The text repeatedly describes the effect as 'damping,' and the reported lower bound on M_LV relies on this sign: with a positive friction coefficient the spectrum would be suppressed and the claimed lower bound would not follow. Please state the assumed sign of c1, reconcile the terminology and the equations, and rerun the analysis if the sign changes.","section":"§2-§3, Eqs. (2.8), (3.4), (3.8)"},{"comment":"The scale factor at horizon entry, a_e, is never computed or specified. The text merely says that 'ae refers to the scale factor at horizon entry,' but a_e^{-2} enters exponentially in Eq. (3.8); for f=10^{-9} Hz in standard cosmology a_e is of order 5×10^{-12}, so a_e^{-2} is of order 4×10^{22}, and an error in a_e shifts the inferred M_LV by a factor proportional to a_e^{-1}. Please give the explicit formula for a_e(k) (e.g., the radiation-entry relation a_e = H0 sqrt(Ω_r)/k) and use it consistently across the fitted band. The statement that the result is 'obtained with f=10^{-9} Hz' also needs clarification: if a_e depends on k, the exponent scales faster than f^2 and the posterior should be driven by the highest frequencies, not the lowest; if a_e is instead fixed, that approximation must be justified.","section":"§3, Eq. (3.4); §4"},{"comment":"The central spectral formulas are asserted without derivation. Eq. (2.6) is stated as the equation of motion of action (2.1), but the variation is not shown; in particular, the claim that the R_ij R^ij term cancels the ∇_k K_ij ∇^k K^ij contribution so that GWs remain luminal is not demonstrated. Likewise, the transfer function decomposition in Eqs. (3.3)-(3.4) and the evaluation of D are taken as given, and the step from Eq. (3.7) to Eq. (3.8) skips the computation of D' and the validity of the approximation T'_GR = k T_GR. Please provide these derivations or give precise references for each step so the modified spectral energy density can be verified.","section":"§2, Eq. (2.6); §3, Eqs. (3.3)-(3.4), (3.7)-(3.8)"},{"comment":"The statistical setup is under-specified. The manuscript states only that PTArcade is used with uniform priors, without saying which likelihood is adopted (full Hellings-Downs correlation vs. common-spectrum process), which pulsar noise models are included, and how the one-sided 68% limit is defined. The marginal posterior for log10 M_LV should be shown directly; if it is truncated by the prior boundary at log10 M_LV = -25, the quoted lower limit may be prior-driven. Please report the full posterior or at least the one-dimensional marginalized distribution and convergence diagnostics.","section":"§4"}],"minor_comments":[{"comment":"There are several typographical errors: 'IPT A2' and 'ML V' appear in Section 5, and the Table 1 caption repeats 'MLV MLV' and 'γ γ'; these should be corrected.","section":"§5, Table 1"},{"comment":"The factor (2πf H0/(c M_LV^2) + 1)^2 is numerically very close to unity for the quoted parameter ranges; stating this explicitly would help readers see that the exponential term dominates the constraint.","section":"§3, Eq. (3.8)"},{"comment":"The caption does not specify the axes or the exact quantity plotted; please add axis labels and clarify whether the theoretical curves are Ω_GW from Eq. (3.8) evaluated at the best-fit parameters.","section":"Figure 2"},{"comment":"Reference [54] (PTArcade) lacks a journal or volume entry, and reference [47] is cited for the construction of the action; please complete the bibliographic details.","section":"References"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I think this one is worth a referee's time, but only after the authors fix the presentation. The new piece is real: first application of this specific Lorentz-violating GW damping model to PTA data, and it reports a bound M_LV > 10^-19 GeV, two orders of magnitude stronger than the LIGO/Virgo bound from inspirals. The spectral energy density for the model in the PTA band is derived, and the parameter estimation with PTArcade is standard.\n\nThe soft spots are in the derivation and reproducibility. The equations of motion and transfer function are asserted with minimal derivation, and the paper relies on refs [37, 50, 51] but does not show enough to check the sign conventions. Most importantly, the exponent in Eq. (3.8) is positive for a_e < 1, so the model amplifies the spectrum, while the text repeatedly calls it damping. That is not cosmetic: the sign decides whether the model suppresses or enhances the signal. The numerical bound depends directly on a_e, the scale factor at horizon entry, which is never stated—the authors say it matters but do not report it. Without a_e, the headline constraint is not reproducible.\n\nOn the frequency dependence, the paper's statement that higher frequency gives a stronger bound is actually consistent with the exponent, since the exponent grows with f^2. So that part is fine. But the absolute scale of the exponent is set by a_e. If my estimate for a typical PTA mode is right (a_e ~ 10^-12 for radiation-dominated entry), the exponent at f = 10^-9 Hz is tiny unless M_LV is many orders of magnitude below 10^-19 GeV, which would make the bound weak. The fact that they get a non-trivial bound suggests they used a specific a_e, and that value needs to be reported and justified.\n\nThe paper is not a breakthrough—10^-19 GeV is an extraordinarily low energy scale, so the practical impact on quantum-gravity model building is modest. But it is a legitimate new constraint, and the idea that PTAs can probe Lorentz violation via amplitude damping is worth taking seriously.\n\nBottom line: send it to a referee, but require the authors to show the derivation, fix the sign description, and state a_e. As written, the central result is unverifiable.","headline":"New PTA constraint on Lorentz-violating GW damping, but the central bound is unreproducible without the reported horizon-entry scale factor and has a sign inconsistency.","tokens_in":8792,"tokens_out":10485,"would_cite":false,"duration_ms":105062,"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":"In a modified-gravity model with extrinsic-curvature derivative terms, pulsar-timing observations of the nanohertz gravitational-wave background imply a Lorentz-violating scale $M_{LV} > 10^{-19}$ GeV at 68% confidence.","keywords":["Lorentz violation","gravitational waves","pulsar timing arrays","stochastic gravitational-wave background","modified gravity","extrinsic curvature","spectral energy density"],"falsifier":"Compute $a_e$ from the standard horizon-entry condition $k = a_e H(a_e)$ and evaluate Eq. (3.8) at $f = 10^{-9}$ Hz for $M_{LV} = 10^{-19}$ GeV. If the exponential factor is not a suppression of the GR spectrum, the reported bound is not a damping bound. A direct numerical integration of the mode equation (2.6) at that frequency would settle which sign is physical.","tokens_in":7599,"feed_emoji":"📡","tokens_out":14759,"duration_ms":140590,"temperature":0.7,"pith_summary":"The paper's central claim is that pulsar-timing observations of the nanohertz gravitational-wave background can directly constrain Lorentz violation in gravity. It adds extrinsic-curvature derivative terms to the gravitational action, making gravitational-wave propagation frequency-dependent through an energy scale $M_{LV}$, and derives the resulting spectral energy density. Fitting this spectrum to two recent pulsar-timing datasets yields $M_{LV} > 10^{-19}$ GeV at 68% confidence, excluding Lorentz-violating gravitational modifications with energy scales below that value and improving on the earlier binary-merger bound by about two orders of magnitude.","feed_headline":"Pulsar timing data push Lorentz-violation scale past 10^-19 GeV","feed_subtitle":"Modified-gravity damping below this energy scale is ruled out by the observed gravitational-wave background.","key_machinery":"The central object is the modified tensor-mode transfer function and the spectral energy density built from it. The added extrinsic-curvature term changes the gravitational-wave mode equation to $h_A'' + (2+\\bar\\nu)\\mathcal{H}h_A' + k^2 h_A = 0$, where $\\bar\\nu\\mathcal{H} = [\\ln(1 + c_1 k^2/a^2)]'$. The load-bearing piece is the closed-form approximation $T = e^{D}T_{\\mathrm{GR}}$ with $D$ as above; it converts the single parameter $M_{LV}$ into a frequency-dependent deformation of $\\Omega_{GW}(f)$, and the fits map the data onto a lower bound for $M_{LV}$.","core_discovery":"The central claim is that adding the term $c_1(\\nabla_k K^{ij}\\nabla^k K_{ij} - R_{ij}R^{ij})$ to the gravitational action, with $c_1 = \\alpha_{\\bar\\nu}/M_{LV}^2$, changes the stochastic gravitational-wave background at pulsar-timing frequencies. The transfer function picks up an exponential factor $e^{2D}$ with $D = -\\frac{1}{2}[\\alpha_{\\bar\\nu}(2\\pi f/(a c M_{LV}))^2]_{a_e}^{a}$, producing the present-day spectral energy density in Eq. (3.8). Fitting that spectrum to two pulsar-timing datasets gives $\\log_{10}(M_{LV}/\\mathrm{GeV}) > -19$ at 68% confidence for both, with best-fit strain amplitudes $\\log_{10}A \\approx -14.13$ and $-14.34$, and spectral indices $\\gamma \\approx 3.22$ and $4.08$. The paper reads this as evidence that pulsar timing arrays probe Lorentz violation more strongly than binary-merger observations.","pith_inferences":["A natural check is to fix $a_e$ from the horizon-entry condition $k = a_e H(a_e)$ and recompute Eq. (3.8); the reported lower bound would shift if this changes the sign or size of the exponential factor.","The same exponential transfer function predicts a departure from a pure power-law background at higher frequencies, so future pulsar-timing data with sensitivity above roughly $10^{-8}$ Hz could confirm or rule out the model independently of the current fit.","The derivation is not tied to nanohertz frequencies; recasting it at millihertz band would extend the bound by several orders of magnitude if the model holds, though the paper does not perform that extrapolation.","If the model is correct, power-law spectral-index templates partially absorb the Lorentz-violating curvature, so joint fits that include the exponential shape should be used when comparing models."],"forward_implications":["Lorentz-violating gravitational modifications with an energy scale below $10^{-19}$ GeV are excluded by current pulsar-timing data, within this model.","The same datasets imply a blue-tilted background spectrum ($\\gamma \\approx 3.2$ and $4.1$), so the observed signal can be accommodated as an inflationary relic with modified propagation.","Because the exponential factor grows with frequency, the bound strengthens toward higher reference frequencies; the paper reports roughly an order-of-magnitude improvement at $f_{yr}$ compared with $10^{-9}$ Hz.","Pulsar timing arrays become a competitive probe of spacetime symmetries, reaching about two orders of magnitude below the binary-merger energy scale."],"supporting_citations":[{"why":"Supplies the 15-year pulsar-timing dataset that provides the observed background spectrum used in the fit.","marker":"[29]"},{"why":"Supplies the second-data-release search for an isotropic gravitational-wave background, the other dataset used in the fit.","marker":"[46]"},{"why":"Sets the earlier binary-merger lower bound of $10^{-21}$ GeV that this analysis improves on.","marker":"[37]"},{"why":"Supplies the derivation of the spectral energy density formula for a stochastic gravitational-wave background used to build Eq. (3.8).","marker":"[50]"},{"why":"Supplies the Hubble constant and radiation density used in the spectral density calculation.","marker":"[52]"},{"why":"Provides the wavenumber-to-frequency conversion used to evaluate the bound at PTA frequencies.","marker":"[53]"},{"why":"Provides the analysis pipeline used to run the parameter estimation and produce the reported posteriors.","marker":"[54]"}],"fun_headline_variants":["Pulsar timing data tighten Lorentz-violation bound past 10^-19 GeV","NANOGrav and IPTA impose sharpest Lorentz-violation scale limit yet","Gravitational-wave background from pulsars pins down Lorentz symmetry","Pulsar timing arrays outshine LIGO in testing Lorentz invariance"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The bound depends on the sign and size of the exponential factor in Eq. (3.8), which in turn depends on the horizon-entry scale factor $a_e$ and the sign convention adopted; if either is wrong, the reported $M_{LV}$ limit changes.","fun_headline_variants_meta":{"raw":{"variants":["Pulsar timing data tighten Lorentz-violation bound past 10^-19 GeV","NANOGrav and IPTA impose sharpest Lorentz-violation scale limit yet","Gravitational-wave background from pulsars pins down Lorentz symmetry","Pulsar timing arrays outshine LIGO in testing Lorentz invariance"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00091,"raw_usage":{"total_tokens":3907,"prompt_tokens":938,"completion_tokens":2969,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":554,"completion_tokens_details":{"reasoning_tokens":2888}},"tokens_in":554,"tokens_out":2969,"duration_ms":23985,"temperature":1.0,"reasoning_tokens":2888,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:03:51.416197+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $a_e$ from the standard horizon-entry condition $k = a_e H(a_e)$ and evaluate Eq. (3.8) at $f = 10^{-9}$ Hz for $M_{LV} = 10^{-19}$ GeV. If the exponential factor is not a suppression of the GR spectrum, the reported bound is not a damping bound. A direct numerical integration of the mode equation (2.6) at that frequency would settle which sign is physical.","supporting_citations":[],"review_version":1}