{"id":"4d5d27a9-6946-478a-8fae-13fa254e3f4b","arxiv_id":"2607.18442","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Global asymptotic shape, not just local curvature, controls tachyonic self-resonant preheating of string moduli, and stochastic light-tower effects mostly smear existing resonance bands.","lead":"This paper studies how the shape of string-theory potentials—flat plateaus versus sloping runaways—controls whether a modulus field can efficiently convert its energy into particles after inflation. It also asks whether the tower of light particles predicted by the swampland distance conjecture changes that efficiency, and finds the effect is mostly a modest smearing of existing resonance bands.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"SDC-tower conclusion rests on an uncomputed noise variance; Sec. 3.2 leaves Var(ξ) free, so 'mainly reshape' is not derived from tower parameters.","rationale":"The paper has two logically distinct results. The deterministic claim—global asymptotics control tachyonic self-resonance—is supported by the semi-analytic gain formula (Eq. 2.11) and by consistent Floquet-map comparisons across blow-up, LVS, and KKLT examples; I do not see a load-bearing flaw there. The stochastic claim is where the SDC tower actually enters, and there the link from the tower Lagrangian to the noise strength in Eq. 3.7 is missing. ξ² is fixed by the χ_n amplitudes, but those amplitudes are never derived; Fig. 13 only establishes the shape of the distribution, not its variance. The paper's own Appendix D states the required validity conditions precisely (Neff≫1, no dominant mode, ϵ_ξ≲1, energy subdominance) but leaves them uncomputed. As a result, the qualitative conclusion that moderate noise reshapes rather than opens a new channel cannot be confirmed or refuted from the paper: any outcome could be accommodated by choosing Var(ξ). This is the same weakest point the Reader identified. It does not undermine the deterministic preheating analysis, but it does mean the paper's second advertised conclusion is not yet established. The appropriate verdict remains CONDITIONAL, pending a concrete derivation of ξ_i² from the coupled equations or a lattice simulation.","tokens_in":21948,"tokens_out":7703,"duration_ms":97334,"concrete_test":"Compute the cycle-averaged ξ_i² of Appendix D for the LVS parameters of Sec. 3.2 by solving the coupled zero-mode + χ_n system (Eqs. 3.5 and 3.9) with χ_n initialized at vacuum (or at zero amplitude), using the actual KK mass spectrum m_n0 and displacement α. Then evaluate Neff and ϵ_ξ using the definitions in Appendix D. If the resulting √Var(ξ/m) lies below the moderate range (0.5–1) used in Figs. 14–15, the claimed band reshaping is negligible; if it lies above the controlled range, the stochastic approximation violates ϵ_ξ≲1 and the free-variance results are not applicable. Either outcome settles whether the SDC-tower reshaping claim is physically realized.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central stochastic claim—that SDC light towers mainly reshape existing resonance bands—is not derived from the tower parameters. In Sec. 3.2, Eq. 3.7 defines ξ² in terms of the tower masses and χ_n amplitudes, but Fig. 13 is used only to establish Gaussianity, and the text then explicitly says the Gaussian result lets the authors 'freely choose the variance of ξ entering our numerical estimates.' No calculation is given for the variance from the LVS/KKLT spectra, the χ_n initial amplitudes, or the cycle-averaging defined in Appendix D. Appendix D lists the required validity conditions (Neff≫1, no single mode dominating, ϵ_ξ≲1, ρ_χ≪ρ_φ) but does not compute any of them. Because the noise amplitude is a free dial, the finding that σ=0.5–1 only smears bands while σ=5 seeds bursts cannot be mapped to a physical regime. Moreover, if the χ_n are initially unexcited, ξ² is negligible and the stochastic correction disappears entirely; the paper does not show that the tower is populated with the amplitudes needed for its noise model. The stochastic conclusion is therefore conditional on an unverified assumption, exactly as the paper itself flags.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies self-resonant preheating of a single string modulus, with two Swampland-motivated themes. First, it argues that the local curvature diagnostic η_V = M_pl^2 |V''/V|, related to the refined de Sitter conjecture, is insufficient to determine tachyonic resonance efficiency; the global shape of the potential determines how long the oscillating condensate dwells in the tachyonic region. Using a semi-analytic tachyonic-gain formula (Eq. 2.11) and Floquet maps, it categorizes plateau-type potentials (blow-up moduli, α-attractor models) as inefficient (γ/H ~ 0.3) and runaway-type potentials (LVS, KKLT) as more efficient (γ/H ~ 1–10). Second, the paper models SDC light KK towers as a stochastic, cycle-to-cycle Gaussian modulation of the Hill equation and finds that moderate noise mainly smears, shifts, and mildly seeds existing resonance bands, rather than opening a robust new reheating channel; energy deposition into the KK sector is argued to be minimal.","tokens_in":22330,"tokens_out":9822,"duration_ms":86030,"significance":"If the deterministic conclusion holds, the paper provides a useful and relatively simple diagnostic — Eq. (2.11) — that connects local curvature, global asymptotics, and preheating efficiency, and it extends the preheating discussion to the Swampland program. The paper is honest about the qualitative nature of the γ/H expansion treatment and about the limitations of the stochastic model; the Floquet maps and semi-analytic comparisons are reproducible in structure. The stochastic conclusion, however, is currently conditional: the variance of the effective noise is not derived from the tower parameters, and the paper's own Appendix D lists the required validity conditions without computing them. The central claim about global asymptotics is supported by the derivation of Eq. (2.11), but the numerical examples do not cleanly separate local and global effects.","major_comments":[{"comment":"The variance of the stochastic noise ξ is never computed from the tower parameters. After Fig. 13 the text states that the Gaussian result allows the authors 'to freely choose the variance of ξ entering our numerical estimates.' This is a free dial, not a derived quantity. Appendix D lists the conditions for the stochastic approximation (N_eff ≫ 1, no single dominant mode, ϵ_ξ ≲ 1, ρ_χ ≪ ρ_φ) but does not evaluate any of them for the LVS/KKLT spectra or for the assumed initial KK amplitudes X_n. Since the paper's abstract claims that SDC towers mainly reshape rather than open a robust channel, that conclusion is not yet established. A calculation or controlled bound on Var(ξ) — from m_n0, X_n, α, and the cycle average in Eq. (D.1) — is needed to connect the stochastic maps to a physical regime.","section":"§3.2, Eq. (3.7), and Appendix D"},{"comment":"The numerical demonstration that global asymptotics, rather than local curvature alone, controls preheating efficiency is not cleanly isolated. The blow-up example has η_V < 1 in the tachyonic region, while KKLT has η_V ≫ 1; the difference in γ/H could therefore be attributed to local curvature alone. Eq. (2.11) analytically shows the additional global factor V(Φ)/V(φ), which is the paper's main point, but the examples pair different local and global data simultaneously. A matched comparison — two potentials with similar local η_V profiles but different asymptotic tails, or a decomposition of Eq. (2.11) into local and global contributions — would make the title's claim more robust.","section":"§2.3–2.5"}],"minor_comments":[{"comment":"The formula as written appears dimensionally inconsistent unless one explicitly sets M_pl = 1. Please state the unit convention, or write the integrand with φ normalized by M_pl.","section":"§2.1, Eq. (2.11)"},{"comment":"The caption says 'a tower of 20 coherently oscillating scalars χ_i with masses n m_n0 e^{-αφ_1/M_pl}, where integers n∈(0,10^4) are sampled uniformly.' It should clarify whether 20 modes are sampled from that range, and what the resulting distribution of masses is.","section":"Fig. 13 caption"},{"comment":"The notation is confusing: ξ^2 is defined in Eq. (3.7), but the text treats ξ as the Gaussian noise (as in Fig. 13). Specify whether the Gaussian variable is ξ or ξ², since the latter is chi-square distributed with one degree of freedom.","section":"§3.2, around Eq. (3.7)"},{"comment":"The conditions N_eff ≫ 1, max_n σ_n^2/Σσ_m^2 ≪ 1, ϵ_ξ ≲ 1, and ρ_χ ≪ ρ_φ are listed but not evaluated. Even an order-of-magnitude estimate for the LVS benchmark would help the reader assess whether the stochastic regime is physically realized.","section":"Appendix D"},{"comment":"Typos: 'Institude' in the first affiliation; 'depeding' in the caption of Fig. 8. Also, the α-attractor section (2.4) largely refers to prior work without showing Floquet maps; this is acceptable but should be stated explicitly.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of JHEP and addresses an interesting question. The deterministic part is largely sound and the semi-analytic diagnostic Eq. (2.11) is valuable. The main weakness is the stochastic section: the variance of the noise is a free parameter, so the abstract's claim about SDC towers is conditional. This is fixable by deriving or bounding the variance from the tower parameters and by checking the Appendix D conditions for the benchmarks. I would also encourage the authors to add a controlled comparison of local vs. global effects, since the current examples conflate them."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Plain talk: the deterministic half of this paper is the real contribution, and it holds up. The claim that tachyonic self-resonance depends on how long the modulus dwells in the tachyonic region—set by global asymptotics, not just local η_V—is well supported by the Floquet maps and the semi-analytic gain formula (Eq. 2.11). The KKLT runaway gives γ/H ~ 10, LVS ~1, blow-up plateau ~0.3. That's a useful categorization, and it's new in the way it ties the refined dS conjecture's η_V bound to dwell-time effects rather than just local curvature.\n\nThe stochastic part (Section 3) is where I part ways. Modeling the SDC tower as Gaussian noise in the Hill equation is a reasonable idea, but the paper never derives the noise variance from the tower parameters. Eq. 3.7 defines ξ² in terms of masses and χ_n amplitudes, but Fig. 13 only establishes Gaussianity, and the text says you can freely choose the variance. That means the headline result—that light states mainly reshape bands rather than open a new channel—is not actually derived; it's an input. If the χ_n are unexcited, ξ² is negligible and the effect disappears. Appendix D lists the conditions under which the stochastic approximation is valid (Neff≫1, no single mode dominating, ε_ξ≲1) but doesn't compute any of them for LVS/KKLT. The paper is honest about this, which I respect, but it doesn't change the fact that Section 3's quantitative conclusion is conditional.\n\nOther soft spots are minor: the γ/H criterion is a qualitative handle, there are no lattice simulations or shipped code/data, and parts of the Floquet analysis repeat [32,41,42,43]. The citation pattern is fine—the stochastic framework is borrowed from [52] and acknowledged.\n\nOverall: worth a serious referee. The deterministic result is clean and useful for a subfield that cares about moduli preheating. Section 3 needs either a calculation of the variance from an actual tower spectrum or an explicit retreat to a toy-model statement. I'd send it to review with that as the main request.","headline":"The global-asymptotics classification of moduli self-resonance is a solid, useful contribution; the stochastic SDC-tower conclusion rests on a free noise variance and is not yet established.","tokens_in":22760,"tokens_out":2370,"would_cite":true,"duration_ms":21983,"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":"Global asymptotics, not local curvature alone, determine whether string modulus preheating is efficient; Swampland light towers mainly reshape existing resonance bands rather than opening a new channel.","keywords":["preheating","string moduli","Swampland conjectures","tachyonic resonance","Floquet analysis","stochastic Hill equation","large volume scenario","KKLT"],"falsifier":"Compute the effective mode number N_eff and the dimensionless variance ratio ϵ_ξ defined in Appendix D directly from the KK masses and initial amplitudes of an explicit compactification; if N_eff ≈ 1 or ϵ_ξ ≫ 1, the Gaussian approximation fails and the claim that towers only mildly seed instabilities is falsified. Alternatively, a lattice simulation that resolves the tower as explicit fields should, for moderate tower masses, reproduce the noiseless Floquet bands with only modest broadening; observing a strong new instability band in a previously stable region would contradict the paper's stoc","tokens_in":21843,"feed_emoji":"🌌","tokens_out":13639,"duration_ms":95675,"temperature":0.7,"pith_summary":"This paper argues that the efficiency of self-resonant preheating of string moduli — the exponential amplification of a modulus's own fluctuations as it oscillates after inflation — is governed by the global shape of its potential, not merely the local curvature data that Swampland conjectures constrain. Two potentials with similar local inflection data can produce parametrically different tachyonic amplification once their asymptotic tails are taken into account, because the tails fix how long the field dwells in the tachyonic region and how it recovers into a stiff regime. In concrete type IIB compactifications, this means runaway potentials like KKLT can support efficient self-resonance (growth rate γ/H ≈ 10), plateau potentials like blow-up moduli are inefficient (γ/H ≈ 0.3), and the LVS volume modulus sits in between. The paper also models the tower of light states predicted by the Swampland Distance Conjecture as a stochastic environment and finds that it mainly smears, shifts, and mildly seeds instabilities in existing resonance bands rather than opening a new reheating channel. A careful reader would care because preheating sets the post-inflationary thermal history and determines whether heavy moduli can be diluted, connecting Swampland constraints to observable early-universe cosmology.","feed_headline":"Runaway moduli preheat at γ/H≈10; plateau moduli stall near 0.3","feed_subtitle":"Far-field shape, not local curvature, decides modulus self-resonance; Swampland towers only smear the bands.","key_machinery":"The central object is the Hill equation for the Fourier modes of the modulus fluctuation, δφ_k, whose time-dependent frequency is set by the oscillating zero mode through m_eff^2(t) = V''(φ̄(t)); the analysis proceeds by Floquet theory, with the growth rate γ extracted from the transfer matrix over an oscillation period. The load-bearing identity is the integrated tachyonic gain per cycle, ln G_k ≈ ∮ dφ̄ √η_V / √(2(V(Φ)/V(φ) − 1)), which shows explicitly that tachyonic amplification depends on both the local curvature diagnostic η_V and the global ratio V(Φ)/V(φ) — that is, on how steeply the potential falls and how long the field dwells in the unstable region. For the tower of light states,","core_discovery":"On its own terms, the paper's central discovery is that the tachyonic branch of the refined de Sitter Conjecture yields a local curvature diagnostic — the second slow-roll parameter η_V = M_pl^2 |V''|/V — that is necessary but not sufficient for efficient self-resonant preheating of a string modulus. The full contribution from tachyonic oscillation depends also on the dwell time inside the tachyonic region and on the recovery to a stiff, up-curving regime, both controlled by the global asymptotics of the potential: plateaus, barriers, and runaways. In the concrete compactifications studied, KKLT runaway potentials give growth rates γ/H ≈ 10, the LVS volume modulus gives γ/H ≈ 1, and plateau-","pith_inferences":["If the stochastic coarse-graining holds beyond the LVS/KKLT examples, the same cycle-to-cycle noise mechanism should apply to any SDC tower — winding modes, string excited states, or axion-like particles — predicting that light towers generically erode sharp resonance wedges in moduli preheating across the landscape.","The runaway-versus-plateau contrast suggests a selection rule the paper does not state: compactifications whose moduli potentials have Dine-Seiberg runaways are more likely to experience efficient self-preheating and avoid the cosmological moduli problem, while plateau-type moduli likely need explicit couplings to other sectors.","The validity conditions in Appendix D could be turned into an explicit test: computing N_eff and ϵ_ξ for a concrete Calabi-Yau KK spectrum would determine whether the Gaussian noise regime is actually realized, and would sharpen or overturn the paper's stochastic conclusion."],"forward_implications":["In Dine-Seiberg runaway compactifications such as KKLT, self-resonant preheating of the volume modulus can be efficient, with growth rates γ/H on the order of 10, making fragmentation and oscillon formation plausible for the oscillating modulus.","In plateau-like blow-up potentials, tachyonic self-resonance is inefficient — γ/H around 0.3 — so reheating in these corners must rely on channels beyond the modulus's own self-interactions.","The tachyonic branch of the refined de Sitter Conjecture, through its lower bound on η_V, favors local curvature that enhances tachyonic gain; Swampland-motivated curvature data thus points toward, but does not guarantee, efficient self-resonance.","For the stochastic treatment, the SDC tower acts mainly as a dephaser: it smears and shifts existing resonance bands and produces at most small growth (γ/H ≲ 0.1) in regions stable in the noiseless map, so a modulus plus a light KK tower is unlikely to preheat the universe efficiently on its own."],"fun_headline_variants":["Swampland dictates preheating: runaway moduli resonate, plateaus stall","Global potential shape, not curvature, decides modulus preheating","Tachyonic curvature is not enough: moduli preheat by global asymptotics","Light-state towers only smear resonance bands, not drive preheating","Runaway moduli preheat fast, plateaus stall: Swampland shapes it"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the tower of light states predicted by the Swampland Distance Conjecture can be coarse-grained into a Gaussian stochastic environment with a freely chosen cycle-to-cycle variance; if a few coherent modes dominate the backreaction, or if the variance cannot be derived from tower parameters, the conclusion that light towers only mildly reshape resonance bands is not established.","fun_headline_variants_meta":{"raw":{"variants":["Swampland dictates preheating: runaway moduli resonate, plateaus stall","Global potential shape, not curvature, decides modulus preheating","Tachyonic curvature is not enough: moduli preheat by global asymptotics","Light-state towers only smear resonance bands, not drive preheating","Runaway moduli preheat fast, plateaus stall: Swampland shapes it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000503,"raw_usage":{"total_tokens":2342,"prompt_tokens":843,"completion_tokens":1499,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":587,"completion_tokens_details":{"reasoning_tokens":1413}},"tokens_in":587,"tokens_out":1499,"duration_ms":14224,"temperature":1.0,"reasoning_tokens":1413,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:23:06.361982+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the effective mode number N_eff and the dimensionless variance ratio ϵ_ξ defined in Appendix D directly from the KK masses and initial amplitudes of an explicit compactification; if N_eff ≈ 1 or ϵ_ξ ≫ 1, the Gaussian approximation fails and the claim that towers only mildly seed instabilities is falsified. Alternatively, a lattice simulation that resolves the tower as explicit fields should, for moderate tower masses, reproduce the noiseless Floquet bands with only modest broadening; observing a strong new instability band in a previously stable region would contradict the paper's stoc","supporting_citations":[],"review_version":1}