{"id":"1604b556-5719-4257-8d4f-1c02ada7fb96","arxiv_id":"2411.13101","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"LLR simulations of Sp(4) pure gauge theory at finite temperature show that the plaquette distribution at the deconfinement critical point has a plateau between the two peaks, and the thermodynamic extrapolations of the specific heat and latent heat disagree, possibly due to mixed-phase effects.","lead":"This lattice paper applies the density-of-states (LLR) method to the deconfinement transition in Sp(4) gauge theory, reporting first thermodynamic-limit extrapolations of the specific heat and plaquette jump at fixed N_t=4. It also presents evidence that the plaquette distribution at the critical point deviates from the standard double-Gaussian shape, suggesting mixed-phase configurations must be included in future extrapolations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The double-Gaussian breakdown claim rests on a 3-sigma intercept discrepancy whose quoted errors the authors themselves suspect are underestimated; LLR convergence in the between-peak region is not demonstrated.","rationale":"The reader's weakest_assumption pointed to the plateau being an LLR artifact; my concern overlaps but emphasizes that even if the plateau is real, the paper's quantitative claim of a thermodynamic-limit discrepancy is not robust because the error budget is self-admittedly incomplete. The plateau itself is theoretically expected for mixed-phase interfacial configurations, so I do not treat its existence as outside consensus; the issue is whether the numerical evidence here demonstrates it. The data and code release is real independent support that makes the proposed test straightforward. I therefore keep the reader's CONDITIONAL verdict unchanged rather than rejecting.","tokens_in":11712,"tokens_out":10714,"duration_ms":117709,"concrete_test":"Re-analyze the released LLR data (Zenodo [66]) with a fully correlated jackknife/bootstrap that propagates autocorrelations of the a_n coefficients and the Delta u_p -> 0 limit through the reconstruction of P_beta(u_p) and the finite-size extrapolations in Fig. 2. Compute the bootstrap p-value for the difference between the two thermodynamic-limit intercepts, and the bootstrap error on the integral of the residual between the LLR distribution and the double-Gaussian fit in the between-peak region. If the intercept difference is no longer significant at the 3-sigma level, or if the residual integral is consistent with zero, the central claim is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central conclusion (Sec. 3) — that mixed-phase configurations 'must be hence included in future analysis' — rests on two pieces of evidence: the flat between-peak plateau in the 4x48^3 plaquette distribution (Fig. 3, right) and the thermodynamic-limit mismatch between (4a^4/6V) C_V^(max) = 5.85(2) x 10^-6 and (Delta <u_p>_beta_CV)^2 = 6.09(7) x 10^-6 predicted by Eq. (8). Both are only as strong as the LLR reconstruction in the exponentially suppressed region between the peaks, where the iterative a_n coefficients are hardest to converge and correlations between intervals are largest. The paper itself flags both threats: in Sec. 2 it notes the C_V fit has chi^2/dof = 5.5 and that the small statistical errors may indicate underestimated correlations, and in Sec. 3 it reports that the iterative procedure for a_n requires more steps on the larger 4x80^3 lattice. If quoted errors were inflated by sqrt(chi^2/dof) ~ 2.3, the intercept discrepancy would drop below 2-sigma; if the between-peak a_n are not fully converged, the plateau could be an LLR artifact rather than a physical mixed-phase contribution. The right panel of Fig. 3 does not quantify the significance of the residual (magenta area), and the double-Gaussian fit excludes the between-peak region. The evidence is therefore intriguing but not yet load-bearing enough to establish a breakdown of double-Gaussian finite-size scaling.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This proceedings contribution applies the LLR density-of-states method to the finite-temperature deconfinement transition in Sp(4) pure gauge theory. On lattices with N_t=4 and N_s=20, 24, 28, 40, 48, the authors reconstruct the plaquette distribution and specific heat near the transition. They report that the plaquette distribution at the critical point deviates from a double-Gaussian form, with a plateau between the two peaks that they attribute to mixed-phase configurations. They also extrapolate the normalized specific-heat peak and the squared plaquette jump to the thermodynamic limit using a quadratic fit in N_t^3/N_s^3; the two extrapolations disagree, and the C_V fit has chi^2/dof=5.5. The paper concludes that mixed-phase contributions must be included in future extrapolations and outlines ongoing work on larger volumes and N_t=5,6.","tokens_in":12067,"tokens_out":6324,"duration_ms":60973,"significance":"Should these findings hold, they would challenge the standard double-Gaussian finite-size-scaling analysis for first-order transitions in a theory of interest for dark-sector gravitational-wave phenomenology, and they would demonstrate that LLR can access physics in the exponentially suppressed region between coexisting-phase peaks. The paper is transparent about its limitations, and it provides data and analysis code in Refs. [66] and [67]. The evidence as presented is, however, preliminary: the central claims rest on a roughly 3-sigma intercept difference that is sensitive to error estimates, and on a between-peak plateau whose statistical significance and method-induced systematics are not yet quantified. The conclusion of Section 3 is correspondingly stronger than the current numerical support.","major_comments":[{"comment":"The claim of a statistically significant discrepancy between (4a^4/6V) C_V^(max) = 5.85(2)x10^-6 and (Delta <u_p>_{beta_CV})^2 = 6.09(7)x10^-6 in the thermodynamic limit is not yet robust. The specific-heat fit has chi^2/dof=5.5, and the authors themselves note that the small statistical errors may reflect underestimated correlations. If the quoted C_V intercept error is inflated by sqrt(chi^2/dof) ~ 2.3, the difference drops from about 3.3 sigma to about 2.9 sigma; a more complete treatment of correlated errors and fit quality is needed before this discrepancy can be presented as evidence against the double-Gaussian relation, Eq. (8). The two extrapolations also derive from the same LLR density of states, so combining their errors in quadrature may overstate the significance.","section":"Sec. 2, Fig. 2 and Eq. (8)"},{"comment":"The between-peak plateau is the direct evidence for mixed-phase configurations, but its significance is not quantified. The figure shows a magenta difference area, but no residual chi^2, no fit parameters for the double Gaussian, and no statement of the fit window or of how the plateau region was excluded. The caption's description of the fit region is also ambiguous ('measurements falling outside the region delimited by the two vertical black lines'). Because the entire conclusion depends on this plateau, the paper should report a quantitative test, e.g., the integrated residual and its statistical error, and demonstrate robustness to the choice of fit range and to the LLR interval size Delta u_p.","section":"Sec. 3, Fig. 3 (right)"},{"comment":"The faithfulness of the LLR reconstruction in the region between the peaks is not demonstrated. This is exactly the region where the density of states is exponentially suppressed, where the iterative determination of the a_n coefficients is hardest, and where the finite interval size Delta u_p can smooth the reconstructed distribution. The paper reports that the 4x80^3 lattice requires more iterations, which underscores the convergence risk, but no convergence check is shown for the 4x48^3 data used in Fig. 3. The plateau could therefore be an artifact of incomplete LLR convergence or of finite Delta u_p rather than a physical mixed-phase contribution; this must be addressed before the conclusion of Section 3 can be accepted.","section":"Sec. 2, Fig. 3 (left)"},{"comment":"The thermodynamic-limit extrapolation uses a quadratic polynomial in N_t^3/N_s^3 without justification or robustness tests. With five lattice sizes and three fit parameters, the fits have little discriminating power, and the choice of the scaling variable and fit order can materially change the intercept. The paper should either derive the expected finite-size form, or show stability of the two intercepts under alternative fit choices (e.g., linear versus quadratic, or dropping the smallest volume), and quote a systematic error from this source.","section":"Sec. 2, Fig. 2 (right)"}],"minor_comments":[{"comment":"The acknowledgments contain a typo: 'ST/R00238X/1The DiRAC Extreme Scaling service' should be split into a proper grant code and sentence start.","section":"Acknowledgments"},{"comment":"The text reports chi^2/N_dof=5.5, but with five data points and a quadratic fit the number of degrees of freedom is two; please state the number of points explicitly and define the fit function.","section":"Sec. 2 after Eq. (8)"},{"comment":"The notation (Delta <u_p>_{beta_CV})^2 is used before it is defined; please define it in a displayed equation, e.g., as the squared difference of the two peak locations at the critical coupling.","section":"Sec. 2, Fig. 2 text"},{"comment":"The manuscript relies heavily on Ref. [64] for the interval-size extrapolation ('as discussed in the appendices of Ref. [64]'); please indicate which appendix is used and what the extrapolated values are before the final fit.","section":"Sec. 2, Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"This is a proceedings contribution, and the authors are appropriately cautious in framing the results as an update and a road map. My recommendation of major revision is driven by the gap between the strength of the conclusion ('must be hence included in future analysis') and the current quantitative evidence. I would be comfortable with acceptance after the authors either add the missing error and convergence analysis or soften the central claim to a conjecture. No scholarly integrity concerns; the paper is transparent about its limitations and provides code and data."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline: this is a short LATTICE2024 proceedings update from the LLR/Symplectic project, and the quantitative results are not new here. The paper says plainly that the extended exposition is in the companion paper Ref. [64] (arXiv:2409.19426) and that this contribution adds a discussion of future work. Cite that one, not this.\n\nWhat the paper does well is frame a genuinely interesting tension. Using the LLR density of states, the authors reconstruct the plaquette distribution at the critical point for Sp(4) and compare two thermodynamic-limit estimators: (4a^4/6V) C_V^max and (Delta <u_p>_beta_CV)^2, which Eq. (8) says should coincide in the infinite-volume limit. They find a discrepancy, 5.85(2) x 10^-6 vs 6.09(7) x 10^-6, and a flat between-peak plateau in the 4x48^3 distribution that a double-Gaussian fit misses. They also flag their own caveats: the specific-heat fit has chi^2/dof = 5.5, and the small statistical errors may mean correlations are underestimated. Data and analysis code for the companion paper are on Zenodo, so the work is reproducible, and the comparison is an external identity test, not a fit to a hoped-for conclusion.\n\nThe soft spots are real but the paper does not overclaim. The plateau's significance is not quantified; the double-Gaussian fit excludes the between-peak region, so part of the visual difference is by construction. The LLR reconstruction in the exponentially suppressed between-peak region is exactly where the a_n coefficients are hardest to converge, and the authors note the 4x80^3 lattice needs more iterations. If the quoted errors are underestimated by about the sqrt(chi^2/dof) factor, the intercept discrepancy falls below 2-sigma. The finite-size scaling is assumed to be quadratic in N_t^3/N_s^3 without derivation. So the evidence is intriguing but not load-bearing yet. The authors say \"some evidence\" and \"intriguing evidence,\" which matches the actual strength.\n\nThis is a proceedings summary, not a full paper. The right reader is someone working on density-of-states methods or finite-temperature transitions in Sp(2N) theories; they should go to Ref. [64] for details and treat this as a bulletin. I would not give it primary citation. That said, it is a fair, honest update, and the underlying methodology is solid. The question—whether mixed-phase contributions break the standard double-Gaussian extrapolation—matters for gravitational-wave predictions. It deserves a serious referee rather than a desk reject, with the expectation that the companion paper carries the technical weight.\n\nRecommendation: send it to review, and judge it as a proceedings update whose claims are appropriately hedged.","headline":"An honest, well-caveated proceedings update that frames an interesting possible breakdown of double-Gaussian scaling in Sp(4), but the quantitative weight is in the companion paper and the plateau evidence is not yet quantified.","tokens_in":12615,"tokens_out":3337,"would_cite":false,"duration_ms":29683,"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":"The paper claims that double-Gaussian fits miss a physical mixed-phase plateau in the Sp(4) deconfinement transition, so the standard thermodynamic-limit relation between specific heat and plaquette jump fails unless this plateau is…","keywords":["Sp(4) gauge theory","deconfinement transition","density of states","LLR method","mixed-phase configurations","finite-size scaling","double-Gaussian approximation","lattice field theory"],"falsifier":"Repeat the LLR analysis on the $4\\times80^3$ lattice and at $N_t=5,6$ while monitoring convergence of the $a_n$ coefficients; if the plateau height decreases with volume or the extrapolated values of $(4a^4/6\\tilde{V})C_V^{\\rm(max)}$ and $(\\Delta\\langle u_p\\rangle_{\\beta_{C_V}})^2$ come into agreement once interval-size and correlation systematics are controlled, the claim of a physical mixed-phase contribution collapses. A cleaner test is the volume scaling of the plateau: an interface contribution should scale with surface area, while a numerical artifact would scale differently or vanish as $\\Delta u_p$ shrinks.","tokens_in":11503,"feed_emoji":"⚛️","tokens_out":8271,"duration_ms":76115,"temperature":0.7,"pith_summary":"This paper is trying to establish that the double-Gaussian description of the first-order deconfinement transition in Sp(4) pure gauge theory is incomplete. Using the linear logarithmic relaxation (LLR) density-of-states method, the authors reconstruct the plaquette distribution at the critical point for five lattice volumes with $N_t = 4$, and find a flat plateau between the two coexisting-phase peaks rather than a vanishing probability. They attribute this plateau to mixed-phase configurations, such as bubbles of one phase inside the other. Extrapolating to the thermodynamic limit, the peak of the specific heat and the square of the plaquette discontinuity, which a standard relation (8) requires to coincide, come out as $(4a^4/6\\tilde{V})C_V^{\\rm(max)} = 5.85(2)\\times10^{-6}$ and $(\\Delta\\langle u_p\\rangle_{\\beta_{C_V}})^2 = 6.09(7)\\times10^{-6}$, disagreeing beyond errors. If correct, the result says that mixed-phase physics must be included in infinite-volume extrapolations, which matters for predicting gravitational-wave signals from strongly coupled dark sectors.","feed_headline":"Double-Gaussian fits miss a phase in Sp(4) deconfinement","feed_subtitle":"Density-of-states data show a flat plateau between peaks, so specific-heat and plaquette-jump extrapolations no longer agree.","key_machinery":"The load-bearing object is the plaquette probability distribution $P_\\beta(u_p)$, reconstructed from the density of states via the LLR method, in which $\\ln\\rho(E)$ is approximated piecewise linearly with coefficients $a_n$ on small energy intervals of width $\\Delta u_p$ in plaquette units. From this distribution the paper computes the specific heat $C_V(\\beta)$ and the plaquette jump $\\Delta\\langle u_p\\rangle_{\\beta_{C_V}}$; the double-Gaussian approximation of $P_\\beta$ and the finite-size-scaling relation (8) between $C_V^{\\rm(max)}$ and $(\\Delta\\langle u_p\\rangle)^2$ are the standard benchmark against which the plateau is detected. The plateau itself is the mechanism that breaks the relation, because a flat inter-peak contribution is exactly what the double-Gaussian benchmark cannot represent.","core_discovery":"On the paper's own terms, the central discovery is that at the critical point of the Sp(4) deconfinement transition the LLR-reconstructed plaquette distribution deviates systematically from a double Gaussian: between the two pure-phase peaks the probability does not vanish but forms a plateau, most visible on the largest lattice $4\\times48^3$. Fitting a double Gaussian to this distribution leaves a residual difference in the inter-peak region that the authors interpret as the thermodynamic contribution of mixed-phase configurations. The same conclusion is supported by the thermodynamic-limit extrapolations: using a quadratic fit in $N_t^3/N_s^3$, the normalised specific-heat peak extrapolates to $5.85(2)\\times10^{-6}$ while the square of the plaquette jump extrapolates to $6.09(7)\\times10^{-6}$, violating relation (8) beyond the quoted errors. The authors conclude that double-Gaussian-based extrapolation is insufficient and that mixed-phase configurations must be included in future analyses.","pith_inferences":["If the plateau is a genuine interface contribution, its height should scale with the interface area $\\propto V^{2/3}$ times an interface tension, which gives a direct volume-scaling test to separate physics from LLR reconstruction artifacts.","The same flat-plateau effect is likely to appear in other $Sp(2N)$ and $SU(N)$ pure gauge theories once density-of-states data reach comparable precision, and would show up as a failure of the specific-heat/plaquette-jump relation (8).","A bias in the latent heat from the double-Gaussian approximation propagates into predictions for the gravitational-wave background from composite dark sectors, most directly into the transition strength and inverse-duration parameters.","A practical extension would be to fit the distribution with a double Gaussian plus a constant plateau term and show that the thermodynamic-limit values of $C_V^{\\rm(max)}$ and $(\\Delta\\langle u_p\\rangle)^2$ then satisfy relation (8); this is a direct test the authors have not yet reported."],"forward_implications":["Standard double-Gaussian finite-size extrapolations are insufficient for accurate transition parameters in Sp(4), and the same caution applies to any theory near a first-order confinement transition.","Future analyses of the latent heat, the interface tension, and the gravitational-wave source parameters of the transition must model the mixed-phase contribution to the plaquette distribution.","The reported reduced chi-square of 5.5 for the specific-heat fit indicates that the very small statistical errors may be underestimating correlations in the LLR data, so the extrapolated central values are not yet final.","Larger volumes and temporal extents ($N_t=5,6$) are required to test the plateau's volume scaling and to take the first continuum limit of this transition."],"supporting_citations":[{"why":"Introduces the linear logarithmic relaxation method used to compute the density of states and reconstruct the plaquette distribution.","marker":"[53]"},{"why":"Companion paper whose appendices document the interval-size extrapolation and whose data underlie the thermodynamic-limit fits reported here.","marker":"[64]"},{"why":"Supplies the finite-size scaling laws for first-order transitions used to interpret the volume dependence of the specific-heat peak.","marker":"[65]"},{"why":"Establishes the group's LLR implementation and analysis pipeline in SU(3), which is adapted here to Sp(4).","marker":"[57]"}],"fun_headline_variants":["Sp(4) deconfinement: double-Gaussian fits fail","Plateau in plaquette distribution breaks Sp(4) fits","Density of states reveals mixed-phase plateau in Sp(4)","Sp(4) transition: specific heat vs plaquette jump mismatch","Sp(4) critical point: mixed-phase plateau breaks Gaussian assumption"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The plateau seen between the two peaks of the LLR-reconstructed plaquette distribution is a genuine physical mixed-phase signal, rather than an artifact of incomplete convergence of the $a_n$ coefficients, of the finite interval size $\\Delta u_p$, of underestimated correlations, or of an incorrect placement of the critical coupling.","fun_headline_variants_meta":{"raw":{"variants":["Sp(4) deconfinement: double-Gaussian fits fail","Plateau in plaquette distribution breaks Sp(4) fits","Density of states reveals mixed-phase plateau in Sp(4)","Sp(4) transition: specific heat vs plaquette jump mismatch","Sp(4) critical point: mixed-phase plateau breaks Gaussian assumption"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000656,"raw_usage":{"total_tokens":3032,"prompt_tokens":1000,"completion_tokens":2032,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":616,"completion_tokens_details":{"reasoning_tokens":1940}},"tokens_in":616,"tokens_out":2032,"duration_ms":12694,"temperature":1.0,"reasoning_tokens":1940,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:50:52.225820+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the LLR analysis on the $4\\times80^3$ lattice and at $N_t=5,6$ while monitoring convergence of the $a_n$ coefficients; if the plateau height decreases with volume or the extrapolated values of $(4a^4/6\\tilde{V})C_V^{\\rm(max)}$ and $(\\Delta\\langle u_p\\rangle_{\\beta_{C_V}})^2$ come into agreement once interval-size and correlation systematics are controlled, the claim of a physical mixed-phase contribution collapses. A cleaner test is the volume scaling of the plateau: an interface contribution should scale with surface area, while a numerical artifact would scale differently or vanish as $\\Delta u_p$ shrinks.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the finite-size scaling laws for first-order transitions used to interpret the volume dependence of the specific-heat peak."}],"review_version":1}