{"id":"cbb8fa81-b79e-49a0-8631-a91966db83b8","arxiv_id":"2502.01238","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A free instanton gas random matrix model predicts a singular Dirac spectral peak and a nonvanishing pion-minus-delta susceptibility in the chiral limit of two-flavor hot QCD.","lead":"A simple random matrix model based on a gas of instantons reproduces the low Dirac eigenvalue spectrum of hot QCD, and predicts how the axial U(1) symmetry behaves when quark masses go to zero. The result suggests that U(1)_A breaking survives in the chiral limit for two light flavors, which matters for understanding the QCD phase diagram.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The decisive weakness is the assumed factorization of the quark determinant into ZMZ and bulk parts; without a test of bulk-ZMZ correlation, the full-QCD chiral-limit prediction in Eq. (16) is not secured.","rationale":"The reader's weakest_assumption identifies both the bulk factorization and the 'eigenvalues always smaller than m' assumption. The latter is internally supported by the paper's exact simulation of the matrix model (which confirms chi_top = m^2 chi_0 without relying on Eq. (13) approximation), so it is less load-bearing. The former, bulk factorization, is an external validity assumption connecting the toy model to full QCD; it is asserted but not demonstrated. This is the point where the chain from quenched data to full-QCD chiral-limit predictions is weakest. The reader's verdict of CONDITIONAL is appropriate: the quenched test is strong, but the full-QCD extension needs independent validation. My concern does not change the verdict; it sharpens the condition that must be met. I also note a possible normalization issue in Eq. (16) (missing 1/V if chi_pi - chi_delta is intended to be intensive), but that affects the quantitative value, not the qualitative nonzero prediction, so it is not the primary load-bearing concern.","tokens_in":15286,"tokens_out":16538,"duration_ms":152312,"concrete_test":"On quenched configurations at T=1.1 T_c (e.g., the same ensembles used for Fig. 2), compute the overlap Dirac eigenvalues and split the spectrum into ZMZ (|lambda| < lambda_cut) and bulk (|lambda| > lambda_cut) at the depleted region. For a range of small quark masses m, compute the bulk part of det(D+m) and test whether its expectation is statistically independent of the ZMZ eigenvalue count and magnitude, e.g., by comparing ensembles reweighted with only the ZMZ determinant against those reweighted with the full determinant (or, if available, against full QCD with two dynamical flavors). If the ZMZ-only reweighting fails to reproduce the full determinant's effect on the ZMZ observables, the factorization assumption is falsified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that chi_pi - chi_delta remains nonzero in the chiral limit for N_f=2 depends on the full-QCD weight being the quenched instanton-gas weight times the ZMZ determinant alone. This requires the assertion in Section 4, after Eq. (9), that 'the contribution of the bulk is not expected to be correlated with that of the ZMZ.' No argument or numerical evidence is given for this non-correlation. Both the ZMZ eigenvalues (set by instanton positions and sizes) and the bulk eigenvalues are determined by the same gauge field, and the bulk determinant is a nontrivial function of that field. If the bulk determinant depends on the instanton configuration, then the weight in Eq. (10) acquires an extra factor that can change the effective instanton density and therefore the m-scaling in Eqs. (15)-(16). The model simulations of the ZMZ alone cannot validate this external assumption. The chiral-limit statement that all ZMZ eigenvalues stay below m is supported within the model by the exact reweighting simulations (Fig. 4), so the bulk factorization is the more load-bearing and untested step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a random matrix model for the near-zero 'zero mode zone' (ZMZ) of the overlap Dirac operator in high-temperature QCD, in which instantons and anti-instantons form a free gas with exponentially small mixing. Fitting the two model parameters (the quenched topological susceptibility and a mixing prefactor) to the distribution of the lowest overlap eigenvalue on a 32^3 x 8 lattice at T = 1.1 T_c, the model predicts the lowest-eigenvalue distribution on a larger volume with no further fitting. For full QCD the dynamical quark determinant is restricted to the ZMZ, assuming the bulk part cancels in expectation values. The model predicts chi(m) = m^{N_f} chi_0 for the topological susceptibility and, in the chiral limit, chi_pi - chi_delta ~ m^{N_f-2} chi_0, which for N_f = 2 remains nonzero even though the topological susceptibility vanishes, implying persistent U(1)_A breaking at high temperature.","tokens_in":15508,"tokens_out":12291,"duration_ms":93425,"significance":"The paper's central claim is striking: if valid, it shows that U(1)_A breaking is not restored in the chiral limit of high-temperature QCD, with the breaking strength set by the quenched topological susceptibility. The quenched part of the paper is genuinely strengthened by the out-of-sample volume prediction (Fig. 3): the model parameters are fixed on one volume and the larger-volume distribution is reproduced without refitting. The mass-dependence predictions in Eqs. (15)-(16) are concrete and falsifiable, and the model is simple enough to simulate in regimes inaccessible to lattice QCD. The main caveat is that the full-QCD predictions depend on an asserted factorization of the quark determinant that is not tested; the manuscript would be considerably strengthened by a numerical test of that assumption or by a clear statement of the conditions under which it holds.","major_comments":[{"comment":"The full-QCD weight in Eq. (10) is obtained by dropping the bulk part of the quark determinant based on the statement that 'the contribution of the bulk is not expected to be correlated with that of the ZMZ.' This factorization is the central load-bearing assumption for the chiral-limit prediction in Eq. (16), yet the paper provides no argument or numerical evidence for it. Since both the ZMZ eigenvalues and the bulk eigenvalues are functionals of the same gauge field, the bulk determinant can in principle depend on the instanton configuration and thereby change the effective instanton density away from the quenched chi_0, which would modify the m-scaling in Eqs. (15)-(16). I recommend adding a concrete check, for example computing on the quenched ensembles of Fig. 3 the bulk determinant reweighting factor conditional on the ZMZ eigenvalue distribution, or comparing the model's predictions with dynamical lattice data at finite m where the spectral peak is resolvable.","section":"Section 4, after Eq. (9)"},{"comment":"The derivation of chi(m) = m^{N_f} chi_0 and of Eqs. (15)-(16) assumes that for arbitrarily small m all ZMZ eigenvalues remain much smaller than m, so that the product in Eq. (13) is essentially unity. The paper's support for this is the heuristic argument that a more dilute instanton gas implies larger separations and hence exponentially smaller splittings; this argument implicitly assumes an infinite volume. For a fixed finite volume, as m -> 0 the Poisson gas has a finite probability of containing zero or one instanton, and the smallest-eigenvalue distribution is not of the dilute-gas form used. Section 7 notes that the order of the thermodynamic and chiral limits matters, but the derivation in Section 5 should state the intended ordering (for instance, m -> 0 after V -> infinity) and justify the uniform validity of |lambda_i| << m in that limit.","section":"Section 5, Eq. (13) and following discussion"}],"minor_comments":[{"comment":"The volume factor V in Eq. (16) is ambiguous: if the susceptibility is intensive (per unit volume), the factor should be absent; if it is extensive, its presence should be explained. The m-dependence is unaffected, but the normalization must be clarified for comparison with lattice data.","section":"Section 7, Eq. (16)"},{"comment":"The simulation data in Fig. 4 are shown without error bars, and the range of masses is not stated; adding error bars and the simulation parameters (volume, number of configurations) would support the claim that the data follow m^2 chi_0 'perfectly'.","section":"Fig. 4"},{"comment":"The power-law fit alpha = -0.770(5) is said to be obtained from the 'common envelope' of finite-volume spectral densities, but the fitting procedure (range of lambda, handling of finite-volume effects, statistical errors) is not described; this makes the quoted error difficult to interpret.","section":"Section 6, Eq. (14)"},{"comment":"The statement that 'our arguments are valid up to arbitrarily high but finite temperatures' is a model-based extrapolation; the paper should explicitly acknowledge that the strength of the predicted effect is set by the quenched topological susceptibility, which falls steeply with temperature, so the signal is expected to be very small at high T.","section":"Conclusions"},{"comment":"The predictions of Eqs. (15)-(16) are not compared with any existing dynamical lattice results at finite m; even a qualitative comparison (e.g., with the JLQCD data cited in Refs. [14-18]) would help calibrate the model and test the factorization assumption.","section":"Section 7, Eqs. (15)-(16)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a proceedings contribution that builds on the author's earlier work (Refs. [30], [39]). The main new claim, the nonzero chi_pi - chi_delta in the chiral limit for N_f = 2, is a model prediction and would be significant if confirmed. My main concern is the lack of any test of the bulk-ZMZ factorization; without such a test, the paper's conclusions remain conditional on an assumption that is not derived from QCD. The editor may wish to consider whether the paper's contribution is best framed as a conjectural model prediction or as a derivation; the current framing ('definite predictions') is somewhat strong given that the full-QCD extension is unvalidated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The quenched part of this paper is the real thing: a two-parameter random matrix model that reproduces the lowest overlap Dirac eigenvalues on one volume and then predicts the distribution on a larger volume without touching the parameters. That is a genuine out-of-sample test, and the agreement at 3.5 fm is strong. The power-law spectral density rho(lambda) ~ lambda^{-0.770(5)} with the volume-scaling envelope is also a concrete, quantitative result. The author is upfront that the instanton-liquid random matrix model is old and that the chiral-limit conclusions largely repeat his PRL; the new pieces are the simplified 3D free-instanton version and the precision of the quenched comparison.\n\nThe soft spot is exactly where the stress-test lands. Moving from quenched to full QCD requires Eq. (9), the factorization of the quark determinant into a zero-mode-zone part and a bulk part, plus the assertion that the bulk is uncorrelated with the ZMZ. No argument or numerical evidence is given for the non-correlation. Both sets of eigenvalues come from the same gauge field, so if the bulk determinant depends on the instanton positions, the effective instanton density changes and the m^{N_f-2} scaling in Eq. (16) is no longer secured. The claim that all ZMZ eigenvalues stay below the quark mass is supported inside the model (Fig. 4), so the bulk-decoupling assumption is the load-bearing, untested step. I also note the main plots have no error bars, no code or data are shipped, and the paper is a proceedings write-up; the full-QCD punchline is already in PRL 132, 131902.\n\nIf the factorization could be tested, say by measuring bulk-ZMZ correlations on small lattices or in the model with a simplified bulk, the conclusion would be much stronger. As it stands, this is a well-posed conjecture backed by a solid quenched model. Lattice practitioners working on U(1)_A should read it, and a referee should focus the author on the factorization assumption. It deserves a serious refereeing because the quenched model is a real contribution and the chiral-limit question is important. I would not rely on Eq. (16) in my own work until the bulk correlation is addressed, but I would cite the quenched model.","headline":"Quenched random matrix model is a genuine success; the full-QCD chiral-limit prediction rests on an untested bulk-decoupling assumption.","tokens_in":16052,"tokens_out":3806,"would_cite":true,"duration_ms":31107,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["12.38.Gc","11.30.Rd"],"model":"deepseek-v4-flash","headline":"The paper argues that U(1)_A breaking survives the chiral limit at any temperature in two-flavor hot QCD, carried by a singular spike in the Dirac spectrum that originates from a free instanton gas.","keywords":["U(1)_A breaking","chiral limit","Dirac spectral density","instanton gas","random matrix model","topological susceptibility","overlap Dirac operator","pion-delta susceptibility"],"falsifier":"A direct lattice calculation with dynamical chiral quarks at fixed temperature just above $T_c$ could decide: if $\\chi_\\pi - \\chi_\\delta$ extrapolates to zero as the quark mass goes to zero at large volume, or if the small-eigenvalue density $\\rho(\\lambda)$ develops a gap or becomes analytic as volume grows, the central claim is refuted. A cheaper check is whether the quenched model's predicted volume dependence of the lowest-eigenvalue distribution survives at larger volumes than those fitted.","tokens_in":15089,"feed_emoji":"🔥","tokens_out":5861,"duration_ms":49616,"temperature":0.7,"pith_summary":"This paper tries to establish that the anomalous U(1)_A axial symmetry of QCD is not restored even in the strict chiral limit at high temperature. The mechanism is a singular peak in the density of small Dirac eigenvalues, caused by the mixing of instanton and anti-instanton zero modes in a nearly free, dilute instanton gas. The paper builds a two-parameter random matrix model that reproduces the quenched overlap Dirac spectrum and can be simulated with dynamical quarks for masses and volumes beyond direct lattice reach. From that model it derives that the pion-minus-delta susceptibility stays nonzero for two flavors in the chiral limit, even though the topological susceptibility vanishes.","feed_headline":"A singular Dirac spike keeps axial symmetry broken at any temperature","feed_subtitle":"A free instanton gas drives pion-delta susceptibility to stay nonzero as quark mass goes to zero.","key_machinery":"The central object is the zero-mode zone (ZMZ) random matrix model, a sparse anti-Hermitian matrix built only from instanton-anti-instanton pairs with off-diagonal elements $c \\exp(-\\pi T r_{ij})$, where $c$ is a fitted prefactor and $r_{ij}$ is the spatial distance between the lumps. The number and locations of instantons and anti-instantons follow independent Poisson distributions with the quenched topological susceptibility as the density; this free-instanton-gas ansatz makes the model tractable with dynamical quarks. Generalizing the Banks-Casher integral to the resulting singular spectral density is what turns the model into quantitative predictions for chiral observables such as the condensate and the pion-delta susceptibility.","core_discovery":"In the chiral limit of two-flavor hot QCD, the difference of the pion and delta susceptibilities is predicted to be nonzero, $\\chi_\\pi - \\chi_\\delta \\sim m^{N_f-2} \\chi_0 V$, because the Dirac spectral density develops an integrable singular power law at zero, $\\rho(\\lambda) \\propto \\lambda^\\alpha$ with $\\alpha = -0.770(5)$ in the quenched case and approaching $-1$ as the chiral limit is taken. This singular spike comes from the would-be zero modes of a free gas of instantons and anti-instantons, whose exponential mixing produces eigenvalues that remain far smaller than the quark mass down to arbitrarily small mass. Consequently the standard Banks-Casher integrals must be generalized, and the axial anomaly continues to affect the pion-delta susceptibility even though the topological susceptibility vanishes in the chiral limit.","pith_inferences":["Extension: if the spike exponent indeed tends to $-1$ in the chiral limit, other spectral sums with higher powers of $\\lambda$ in the numerator may develop logarithmic or divergent behavior, so different observables could show different apparent restoration temperatures.","Extension: the model implies a concrete scaling test for direct lattice simulations: at fixed small quark mass, the pion-delta susceptibility difference should grow with volume in the regime where the spike forms.","Extension: the mechanism ties the persistence of U(1)_A breaking to the temperature dependence of the quenched topological susceptibility, which also drives axion physics; if correct, the high-temperature axion mass and the U(1)_A-breaking signal would share the same suppression factor.","Extension: tightly bound instanton-anti-instanton pairs, which the model predicts alongside the free gas, would contribute eigenvalues that stay away from the singular spike and therefore would not alter the chiral-limit predictions for spike-dominated quantities."],"forward_implications":["For two light flavors, $\\chi_\\pi - \\chi_\\delta$ remains nonzero in the chiral limit at any finite temperature above $T_c$, so U(1)_A breaking does not disappear.","The chiral condensate in the high-temperature phase vanishes as $m^{N_f-1} \\chi_0 V$ for $N_f > 1$, consistent with restoration of the non-singlet chiral symmetry.","Taking the thermodynamic limit first is essential: in a finite volume the singularity is regulated and the chiral-limit pion-delta susceptibility difference vanishes.","Because the magnitude of the effect is set by the quenched topological susceptibility, it becomes small at high temperature but never exactly zero, so the phenomenon is strongest just above $T_c$.","Direct lattice observation of the effect requires a chiral Dirac operator for both sea and valence quarks, plus volumes large enough to contain several instantons and anti-instantons."],"supporting_citations":[{"why":"Gives the Banks-Casher formula whose generalization yields the chiral-limit predictions for the condensate and susceptibilities.","marker":"[1]"},{"why":"Supplies the first quenched observation of the spectral spike at zero.","marker":"[3]"},{"why":"Shows the spike is not a quenched or coarse-lattice artifact and argues it becomes singular in the thermodynamic limit.","marker":"[4]"},{"why":"Establishes that an analytic or gapped spectral density would hide U(1)_A breaking in scalar and pseudoscalar correlators, making the singularity necessary.","marker":"[20]"},{"why":"Provides the instanton zero-mode mixing picture on which the random matrix model is built.","marker":"[22]"},{"why":"Supports the free instanton gas picture with Poisson-distributed topological objects in the high-temperature phase.","marker":"[30]"},{"why":"Found a singular spectral density from instantons in an earlier model, providing precedent for the power-law spike.","marker":"[32]"}],"fun_headline_variants":["Instantons make Dirac spike singular, keeping U(1)A broken","U(1)A breaking survives chiral limit via singular Dirac spectrum","Hot QCD: instantons create singular Dirac peak that breaks axial symmetry","Axial symmetry remains broken: instantons cause singular zero-mode peak","Dirac spike from instanton gas prevents chiral restoration"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument relies on the quark determinant splitting cleanly into an instanton-zero-mode part and an uncorrelated bulk part that cancels in expectations, and on every zero-mode-zone eigenvalue staying much smaller than the quark mass all the way down to the chiral limit.","fun_headline_variants_meta":{"raw":{"variants":["Instantons make Dirac spike singular, keeping U(1)A broken","U(1)A breaking survives chiral limit via singular Dirac spectrum","Hot QCD: instantons create singular Dirac peak that breaks axial symmetry","Axial symmetry remains broken: instantons cause singular zero-mode peak","Dirac spike from instanton gas prevents chiral restoration"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000297,"raw_usage":{"total_tokens":1683,"prompt_tokens":866,"completion_tokens":817,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":729}},"tokens_in":482,"tokens_out":817,"duration_ms":8444,"temperature":1.0,"reasoning_tokens":729,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T15:56:23.028844+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct lattice calculation with dynamical chiral quarks at fixed temperature just above $T_c$ could decide: if $\\chi_\\pi - \\chi_\\delta$ extrapolates to zero as the quark mass goes to zero at large volume, or if the small-eigenvalue density $\\rho(\\lambda)$ develops a gap or becomes analytic as volume grows, the central claim is refuted. A cheaper check is whether the quenched model's predicted volume dependence of the lowest-eigenvalue distribution survives at larger volumes than those fitted.","supporting_citations":[{"cited_title":"Alexandru and I","cited_arxiv_id":null,"evidence_quote":"Shows the spike is not a quenched or coarse-lattice artifact and argues it becomes singular in the thermodynamic limit."},{"cited_title":"Sharan et al","cited_arxiv_id":null,"evidence_quote":"Found a singular spectral density from instantons in an earlier model, providing precedent for the power-law spike."}],"review_version":1}