{"id":"b378d5df-291a-48f1-a1fc-c94139ee1642","arxiv_id":"2509.12414","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Flavour deconstruction phase transitions can generate detectable gravitational waves, but their spectra typically peak above the millihertz range, making LISA detection possible yet not guaranteed.","lead":"This paper computes the gravitational wave signals that two generic TeV-scale flavour deconstruction models would produce during a first-order phase transition. It finds that such signals are possible, but often peak above LISA's millihertz band, so a LISA detection is not guaranteed and mid-band observatories may be needed.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"One-loop effective potential without daisy resummation calls the quantitative SNR and peak-frequency claims into question for the strong-coupling, T_n ~ 0.3v benchmark.","rationale":"The reader's weakest assumption identified the one-loop effective potential without daisy resummation as the key fragile input. My analysis agrees: this is the most load-bearing concern because it directly enters every quantitative prediction that supports the central claim—SNR values, peak frequencies, and the natural-versus-tuned distinction between the non-Abelian and Abelian benchmarks. The paper is transparent about the omission and even labels the impact 'non-negligible,' but the justification for expecting no qualitative change is a heuristic (φ ∼ T_n near the barrier) that actually marks the regime where the ring contribution is largest. The concrete test I propose would settle the question by recomputing a few representative points with a daisy-resummed potential. If the changes are small, the paper's conclusions are robust; if large, the quantitative claims would need revision, though the qualitative statement that FD models can produce signals in the LISA/mid-band range could still hold. Because the paper explicitly acknowledges this limitation and the reader's verdict is already CONDITIONAL, my stress test does not change the recommended verdict. I do not see a more fundamental internal inconsistency: the model setup, the use of matching conditions, and the general GW formalism are standard. The numerical incompleteness of the U(1) scan is a separate issue, but it affects a secondary conclusion and is already noted in the paper.","tokens_in":13575,"tokens_out":4861,"duration_ms":55686,"concrete_test":"For the representative benchmark point with the largest LISA SNR in Fig. 1 (near g4 ≈ 1.5, λ ≈ 10^-2, v = 1 TeV), recompute the phase transition and GW spectrum using a daisy-resummed effective potential. Use an established scheme such as Arnold–Espinosa or Parwani, or better, the dimensionally reduced EFT via DRalgo and PT2GWFinder, which the paper itself cites. Compare T_n, α, β/H, the peak frequency, and the LISA SNR against the one-loop values reported. If the SNR changes by more than ~30% or T_n shifts by more than ~10%, the paper's quantitative predictions are not reliable and the verdict should remain conditional. Repeat for two additional points: one at the edge of the detectable region in the non-Abelian benchmark and one in the Abelian U(1) benchmark where the solver succeeds, to test whether the qualitative picture (detectable signals, peaks above millihertz) survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's quantitative results—SNR contours in Fig. 1 and peak frequencies in Fig. 3—are derived from a one-loop effective potential (Eqs. 9–12) that omits daisy resummation. The authors explicitly note this can affect the potential for φ ≲ T_n around the PT, and Footnote 3 argues that the impact should be moderate because T_n ≈ 0.3v and φ ∼ T_n lies just around the barrier location. This reasoning is not reassuring: when φ ∼ T_n and g~O(1), the gauge-boson mass m_V ∼ g_V φ is comparable to T, so the ring (daisy) contribution, which replaces m_V^2 by m_V^2 + Π(T) with Π ∼ g_eff^2 T^2, is a same-order correction, not a small one. Daisy resummation modifies the cubic term in the effective potential—the very term that generates the barrier—and therefore can significantly shift T_n (Eq. 13), α (Eq. 14), and β/H (Eq. 15). Since the GW amplitude scales roughly as (α/(1+α))^2 and the peak frequency f_sw is proportional to T_n (Eq. 16e), these shifts directly affect the predicted detectability and the central statement that signals peak above the millihertz range. A change in T_n from 0.3v to 0.5v, or the reverse, moves f_sw by tens of percent, which in the LISA band is the difference between the peak-sensitivity region and the high-frequency tail. Thus, the specific SNR regions and the 'natural versus tuned' distinction for the U(1) benchmark are not robust until this systematic is quantified. The qualitative conclusion that FD scenarios can produce detectable gravitational waves may survive, but the paper's own Footnote 3 does not settle the quantitative impact without a calculation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies gravitational-wave signals from first-order phase transitions in flavour-deconstruction (FD) models. It reduces the relevant scalar sector to a single light field φ with tree-level potential (8), supplemented by one-loop Coleman-Weinberg and thermal vector-boson contributions (Eqs. 9–12). The tunnelling action is computed with AnyBubble, yielding the nucleation temperature, transition strength, and inverse duration (Eqs. 13–15), which are then converted into GW spectra and LISA SNR (Eqs. 16–17). Two benchmarks are analysed: a non-Abelian SU(4)×SU(3)×U(1) transition and an Abelian U(1) transition. The central claim is that FD models can produce detectable GWs in specific parameter regions—natural for the non-Abelian case with g4≈1.5 and small λ, finely tuned for the Abelian case—and that the spectral peaks tend to lie slightly above the millihertz LISA band.","tokens_in":13963,"tokens_out":6035,"duration_ms":73340,"significance":"If the quantitative predictions hold, this is one of the first systematic scans of GW signatures from generic FD models and it highlights the role of gauge-coupling matching conditions, which previous studies did not emphasize. The paper is commendably explicit about its limitations: no daisy resummation, v_w=1, g*=200, and a one-loop effective potential. It also uses numerically evaluated thermal functions rather than a high-temperature expansion, and relies on the public code AnyBubble. The qualitative picture—FD provides a plausible target for LISA and mid-band observatories, with a natural non-Abelian region and a tuned Abelian region—is credible. However, the quantitative SNR contours, peak frequencies, and the natural-versus-tuned distinction are not fully controlled until the listed higher-order systematics are quantified or at least bracketed.","major_comments":[{"comment":"The omission of daisy resummation is load-bearing. For the benchmark points with g4~O(1) and T_n≈0.3v, the vector-boson mass m_V≈g_V φ is comparable to T when φ~T_n, so the ring correction replacing m_V^2 by m_V^2+Π(T), with Π∼g_eff^2 T^2, is a same-order effect rather than a small correction. Daisy resummation modifies the cubic term that generates the barrier, which directly shifts T_n in Eq. (13), α in Eq. (14), β/H in Eq. (15), and hence the peak frequency f_sw∝T_n in Eq. (16e). The argument in Footnote 3 that the impact should be moderate because φ∼T_n lies near the barrier is not a substitute for a calculation; a shift of T_n from 0.3v to 0.5v alone changes f_sw by tens of percent and moves signals within the LISA band. I request either a daisy-resummed/dimensionally reduced calculation or a systematic estimate of the sensitivity of the central results (SNR contours and the natural","section":"Noemi Fabri, Gino Isidori, Davide Racco"},{"comment":"The conclusion that the U(1) benchmark is finely tuned and therefore less likely relies on the one-loop effective potential in a regime where T_n/v can be as low as 0.1 and where the numerical solver is unstable (grey points in Fig. 5). In this regime the effective quartic is dominated by loop corrections, and for g4~O(1) with φ∼T_n the one-loop expansion in the gauge coupling is not parametrically controlled. The rapid variation of T_n/v with parameters seen in the central panel could be in part a one-loop artifact. Without higher-order control or a scan of the RG-scale dependence, the 'natural versus tuned' distinction for the Abelian benchmark is not yet robust.","section":"Noemi Fabri, Gino Isidori, Davide Racco"},{"comment":"The LISA SNR in Fig. 1 is computed with v_w=1, which the authors acknowledge overestimates the GW amplitude. Since the SNR contours are used to quantify detectability, a realistic v_w<1 (or at least a rescaling by a range of v_w values) is needed to judge whether the non-Abelian benchmark genuinely reaches LISA sensitivities in natural regions. I am not asking for a full out-of-equilibrium calculation, but a simple parametric scan over v_w∈[0.4,1] would indicate how robust the 'detectable at LISA' claim is.","section":"Noemi Fabri, Gino Isidori, Davide Racco"}],"minor_comments":[{"comment":"The multiplicity c_L is introduced for the vector leptoquark, but the vector is denoted U. Using c_U would avoid confusion.","section":"Noemi Fabri, Gino Isidori, Davide Racco"},{"comment":"The text says 'the most right panel in Fig. 5'; should be 'the rightmost panel'.","section":"Noemi Fabri, Gino Isidori, Davide Racco"},{"comment":"Typo: 'the peak of the GW emission tend to be a frequencies' should read 'tend to be at frequencies'.","section":"Noemi Fabri, Gino Isidori, Davide Racco"},{"comment":"The right panel's vertical axis is not labelled in the text; please state explicitly that it is Ω_GW h² and specify the units.","section":"Noemi Fabri, Gino Isidori, Davide Racco"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid phenomenological study with a clear central claim and unusually explicit caveats. The main reason for major revision, rather than acceptance, is that the one-loop effective potential without daisy resummation is not a controlled approximation in the regime g4~O(1), T_n~0.3v, and the quantitative SNR/frequency statements depend precisely on that calculation. The requested changes are within the scope of the manuscript and would substantially increase the reliability of the conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi X,\n\nQuick take on 2509.12414. This is a well-organized phenomenological study that does exactly what it claims: it maps the GW signal from the last SSB step of two generic flavour deconstruction benchmarks, the non-abelian SU(4)xSU(3)xU(1)' and the abelian U(1)_{B-L} x U(1)'. The main qualitative result—strong transitions occur naturally in the non-abelian case, only by tuning in the abelian case, and the resulting peaks typically sit above LISA's best band, favouring mid-band detectors—is plausible and clearly argued. The reader's conditional verdict is about right.\n\nWhat's genuinely new: relative to the earlier PS3 study [9], this paper identifies the gauge-coupling matching condition as a controlling factor, shows that the combination sqrt(g3^2+g4^2) is minimized around g4~1.5, and documents that strong GW production is not generic across FD parameter space. The abelian versus non-abelian comparison is useful. The paper uses AnyBubble, states its simplifications, and does not overclaim.\n\nThe main soft spot is the omission of daisy resummation. The stress-test note is right: for the benchmark points (g4~O(1), T_n~0.3v, phi~T at the barrier), the ring correction is a same-order effect, not a small perturbation. It modifies the cubic term that generates the barrier, so T_n, alpha, and beta/H can all shift. Since f_sw is proportional to T_n, a few tens of percent shift in T_n moves the peak by tens of percent—enough to matter for LISA SNR. The paper's Footnote 3 says this is 'likely to have a non-negligible impact' but expects no change to the overall picture; that expectation is not backed by a calculation. So the specific SNR contours in Figs. 1-3 and the natural-versus-tuned distinction for the U(1) benchmark should be treated as indicative, not precise. The other simplifications (v_w=1, g* fixed at 200, incomplete U(1) scan) are minor, though the grey areas in Fig. 5 do reduce confidence in the abelian conclusion. No code or data files are released, which would help future checks.\n\nThat said, the qualitative conclusions probably survive. The non-abelian model's large multiplicity of broken gauge bosons naturally creates a strong loop-induced barrier; the abelian model needs cancellation. And the peak frequency above mHz is a simple scaling from T_n~0.3v with v=1-3 TeV, unlikely to be overturned by daisy corrections at the tens-of-percent level. So this is not fatally flawed; it is a useful mapping that needs a second pass on theoretical uncertainties.\n\nWho is this for: GW phenomenologists, FD model builders, and people planning mid-band experiments. It deserves a serious referee, not a desk reject. I'd suggest asking the authors to add a daisy-resummed analysis or at least quantify the ring correction, release the numerical data, and complete the U(1) scan.\n\nRegards,","headline":"A clean mapping of GW signals from flavour deconstruction, with the right qualitative message; the quantitative SNR contours need a daisy-resummed check but the paper deserves a serious referee.","tokens_in":14537,"tokens_out":4530,"would_cite":true,"duration_ms":44156,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Flavour deconstruction models can make detectable gravitational waves, but their TeV-scale signals typically peak just above the millihertz band, so LISA may see them only in part of the parameter space while mid-band observatories become t","keywords":["flavour deconstruction","gravitational waves","first-order phase transition","TeV scale","LISA","gauge coupling matching","link fields","effective potential"],"falsifier":"Compute the two-loop, resummed (dimensionally reduced) effective potential for a representative flavour-deconstruction benchmark point such as g4=1.5, λ around 0.01, v=1 TeV, and re-evaluate the nucleation temperature and barrier height; if the barrier disappears or the transition becomes second-order, the predicted LISA signal-to-noise ratio and the 'possible but not guaranteed' conclusion would be overturned. Alternatively, a sensitive mid-band observatory seeing no signal in the frequency range predicted for T_n≈0.3v would disfavour strong first-order transitions in these models.","tokens_in":13412,"feed_emoji":"🌊","tokens_out":5298,"duration_ms":59065,"temperature":0.7,"pith_summary":"This paper asks whether flavour deconstruction—a family of beyond-Standard-Model theories that explains fermion mass hierarchies by breaking gauge symmetries along flavour directions—leaves a gravitational-wave imprint. The authors find that the TeV-scale link-field sector common to these models can drive strong first-order phase transitions, and that two effective couplings control the resulting gravitational-wave spectrum: the quartic coupling of the light scalar and the gauge-coupling combination fixed by matching conditions. The signals, however, typically peak at frequencies slightly above the millihertz range: LISA may detect them in natural regions of the non-Abelian benchmark, but they are better targeted by mid-band proposals. This converts flavour physics into a concrete observational programme for the next generation of gravitational-wave observatories.","feed_headline":"Flavour-deconstruction waves peak above LISA's band","feed_subtitle":"A first-order TeV-scale phase transition in these models could still be seen by mid-band observatories.","key_machinery":"The argument rests on the one-loop thermal effective potential of the light scalar field that acquires the TeV-scale vacuum expectation value, built from the tree-level potential, the Coleman-Weinberg potential, and the thermal contributions of the heavy gauge bosons. The heavy-vector multiplicities and masses—coloron, vector leptoquark, and Z′—dominate the thermal potential and set the transition strength. The gauge-coupling matching condition g4^{-2}+g3^{-2}=g_s^{-2} is the central identity: it sets the minimum possible value of sqrt(g3^2+g4^2), and therefore fixes the coupling region where the gravitational-wave signal is largest. Tunnelling is treated semiclassically, and the spectrum is","core_discovery":"The central claim is that flavour deconstruction models generically predict strong first-order phase transitions at the TeV scale, sourced by the link fields that deconstruct the flavour gauge group, and these transitions emit gravitational waves with an amplitude large enough to be detectable by planned observatories. The paper isolates the two parameters that govern the signal: λ, the effective quartic coupling of the light singlet field φ, and the gauge-coupling combination sqrt(g3^2+g4^2), whose minimum is fixed by the SM matching relation g4^{-2}+g3^{-2}=g_s^{-2}. Because that minimum occurs at g4≈1.5, the strongest gravitational-wave emission arises at O(1) gauge couplings, which are e","pith_inferences":["Inference: The paper's logic implies that a null result at LISA would not disfavour flavour deconstruction; the signal may simply sit in the mid-band window. Mid-band detectors therefore become the decisive test, and their sensitivity curves deserve modelling as detailed as LISA's.","Inference: The same matching-condition mechanism should apply to any deconstruction chain whose last step lands on the SM gauge group: the minimal mass of the dominant vector boson is set by the SM gauge coupling at the transition scale, which may explain why TeV-scale cascades generically have their gravitational-wave peak pushed above the millihertz range.","Inference: A testable extension is to recompute the phase transition with a dimensionally-reduced, resummed effective potential; if the barrier shifts, the g4≈1.5 window and the 'LISA possible but not guaranteed' conclusion would move accordingly.","Inference: The qualitative contrast between the Abelian (fine-tuned) and non-Abelian (natural) benchmarks could be used as a model discriminator once gravitational-wave data are combined with collider bounds on the new gauge bosons."],"forward_implications":["In the non-Abelian flavour-deconstruction benchmark, natural O(1) gauge couplings produce strong first-order phase transitions with potentially detectable signals at LISA.","The peak frequency of the gravitational-wave spectrum in these models typically lies just above the millihertz range, making mid-band observatories more promising than LISA alone.","In the Abelian (U(1)) variant, a detectable signal requires fine-tuned small quartic couplings, so a gravitational-wave detection would favour non-Abelian flavour-deconstruction structures.","If a signal is seen, gravitational-wave data alone will not cleanly identify the underlying model; combined collider searches for TeV-scale vector leptoquarks and Z′ bosons would be needed to break the degeneracy.","The gauge-coupling matching relation is a new controlling input: at fixed TeV scale, it fixes the minimum of the effective gauge-coupling combination, so detectability is tied to the SM QCD coupling, not a free parameter."],"fun_headline_variants":["Flavour deconstruction emits GWs beyond LISA's sweet spot","Flavour deconstruction waves: too high for LISA, just right for mid-band","LISA may miss flavour-deconstruction waves; mid-band won't","TeV-scale phase transition: gravitational waves for mid-band detectors"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The one-loop effective potential without daisy resummation is assumed to describe the phase-transition barrier reliably for gauge couplings of order one and nucleation temperatures near 0.3v; if higher-order corrections substantially change the barrier, the predicted signal-to-noise regions and the natural-versus-tuned distinction would shift.","fun_headline_variants_meta":{"raw":{"variants":["Flavour deconstruction emits GWs beyond LISA's sweet spot","Flavour deconstruction waves: too high for LISA, just right for mid-band","LISA may miss flavour-deconstruction waves; mid-band won't","TeV-scale phase transition: gravitational waves for mid-band detectors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000738,"raw_usage":{"total_tokens":3096,"prompt_tokens":667,"completion_tokens":2429,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":411,"completion_tokens_details":{"reasoning_tokens":2363}},"tokens_in":411,"tokens_out":2429,"duration_ms":20190,"temperature":1.0,"reasoning_tokens":2363,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T16:36:49.544689+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the two-loop, resummed (dimensionally reduced) effective potential for a representative flavour-deconstruction benchmark point such as g4=1.5, λ around 0.01, v=1 TeV, and re-evaluate the nucleation temperature and barrier height; if the barrier disappears or the transition becomes second-order, the predicted LISA signal-to-noise ratio and the 'possible but not guaranteed' conclusion would be overturned. Alternatively, a sensitive mid-band observatory seeing no signal in the frequency range predicted for T_n≈0.3v would disfavour strong first-order transitions in these models.","supporting_citations":[],"review_version":1}