{"id":"54e7558a-e75a-4bcc-810a-c2eeb0a640f0","arxiv_id":"2510.24403","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A density-of-states counterexample shows that one observable moment vanishing as m^k0 does not imply all moments vanish the same way, breaking a key step in Aoki-Fukaya-Taniguchi.","lead":"This comment argues that a 2012 proof by Aoki, Fukaya, and Taniguchi contains an unjustified step connecting the mass scaling of one positive gauge-field observable to all such observables. If correct, the proof's conclusions that the Dirac spectral density and topological susceptibility vanish at small quark mass in the chirally symmetric phase need reassessment.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The counterexamples target l<l0, but the comment never shows that the l0 used in AFT's Sec. III G exceeds 1; if it is 1, the attack does not engage.","rationale":"The comment's core logical point is sound as a criticism of the general inference: in an arbitrary probability space, one vanishing moment of a positive random variable does not force all moments to vanish at the same rate, and the Gaussian/delta examples demonstrate this cleanly. The m^2-analyticity assumption does not by itself rescue AFT, since the example pO(x,m)=δ(x-m^2) gives moments m^{2l}, which are analytic in m^2; this is an important independent check and the comment does not seriously rely on non-analytic behavior. However, the comment's conclusion that AFT's proof is invalid depends on the claim that the false inference is actually used in Sec. III G in a case where l<l0 matters. The comment quotes the general statement from Ref. [1] but does not identify the specific l0 in the application. If AFT's l0 is 1, the counterexamples do not touch the application, and the proof could be repaired by a boundedness argument. The reader flagged the measure-theoretic heuristic as the weakest assumption; that is related but not identical. The promised additional note is also absent from the provided text, which weakens verifiability but is not the main mathematical issue. The right disposition is therefore conditional: the comment's critique stands only if the relevant AFT step has l0≥2 and if the counterexample distributions can be realized (or are only meant as abstract refutations of a stated general lemma, in which case the comment should say so more precisely).","tokens_in":5577,"tokens_out":20735,"duration_ms":200451,"concrete_test":"Extract from Ref. [1], Sec. III G around Eq. (92), the positive observable O, the integer l0 for which Eq. (1) is established, and the moments l for which Eq. (25) is invoked. If l0=1, the counterexamples (which require l<l0) do not refute the step actually used. If l0≥2, additionally check that the density-of-states family pO(x,m)=δ(x-m^2) (all moments m^{2l}) is compatible with the stated m^2-analyticity and with a finite-volume realization of O as a mass-independent gauge-field observable.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Every counterexample in §f depends on taking l<l0: the Gaussian result (8) has <O^l>/<O^{l0}> ∝ x0^{l-l0} or σ^{l-l0}, and the delta example chooses l0≥2. The comment asserts, but does not verify, that the implication (1)->(3) is crucially used in Ref. [1], Sec. III G after Eq. (92) with a case where this matters. If AFT's application is to an intensive positive observable with l0=1, then for bounded O the inference <O^l>=O(m^{k0}) for l≥1 follows from <O>=O(m^{k0}) by O^l≤C^{l-1}O, and the counterexamples are irrelevant. The comment also explicitly waives rigorous construction of the infinite-volume measures and density-of-states functions (§d), so it is not established that δ(x-x0(m)) or the truncated Gaussian is the distribution of any local gauge-invariant observable under the QCD measure. The mathematical objection to the general lemma is valid; the transfer to AFT's specific proof is the insecure step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript is a Comment on Aoki, Fukaya, and Taniguchi (AFT, Phys. Rev. D 86, 114512), which claims that in the chirally symmetric phase of two-flavor QCD the Dirac spectral density at the origin and the topological susceptibility vanish identically for sufficiently small nonzero quark mass. The Comment isolates a step in AFT's proof in which vanishing of the expectation value of O(A)^{l0} for a positive gauge-field observable O and some l0 is used to infer vanishing of <O(A)^l> for all l. The author argues that Eq. (2) does not follow from Eq. (1), that Eq. (3) does not follow generally from Eq. (2), and supports this with a heuristic measure-support construction in Sec. d and with counterexamples based on truncated Gaussian and delta density-of-states functions in Sec. f. The paper concludes that AFT's results for rho(0;m) and chi_t should be reassessed. The abstract also promises an additional note refuting objections raised by Aoki and Fukaya in their reply, but no such note appears in the submitted text.","tokens_in":5741,"tokens_out":10326,"duration_ms":98776,"significance":"If the objection lands, the Comment identifies a genuine logical gap in an influential proof and would change the evidentiary status of the AFT conclusion that U(1)_A is effectively restored in the chiral limit at high temperature. The paper is valuable for making the l-dependence issue explicit, and it provides clean analytic counterexamples to the general moment-inference lemma. It is also free of circularity and does not smuggle the target conclusion into its assumptions. The force of the Comment, however, is conditional on showing that the contested inference is actually used in AFT with l0 > 1, and on supplying the promised additional note; both points are currently unresolved.","major_comments":[{"comment":"The Comment asserts that the implication (1)->(3) is 'crucially used' after Eq. (92) of Ref. [1], but it never identifies the exponent l0 in that application. The counterexamples in Sec. f produce a failure only for powers l < l0: for instance, Eq. (8) gives <O^l>/<O^{l0}> proportional to x0(m)^{l-l0} or sigma(m)^{l-l0}, which blows up only when l < l0. If the actual application in AFT uses l0 = 1, then for a bounded positive observable O the inference is trivially valid because <O^l> <= C^{l-1}<O> for all l >= 1, and the counterexamples are irrelevant. The manuscript must quote the relevant observable and l0 from Ref. [1] and demonstrate that l0 >= 2; otherwise the central conclusion that AFT's proof collapses is not established.","section":"Sec. g, paragraph after Eq. (92)"},{"comment":"The abstract promises an additional note refuting the objections raised by S. Aoki and H. Fukaya in their reply, and asserts that their claim of having found a mistake in the author's arguments is baseless. No such note appears in the submitted text; the paper ends with Sec. g and the acknowledgments. This is an unfulfilled substantive promise. Either the additional note must be included, or the abstract must be amended to remove the unsupported claim.","section":"Abstract, final two sentences"},{"comment":"The counterexamples are constructed from arbitrary probability densities p_O(x,m) that are assumed to be the density-of-states functions of a positive gauge-field observable under the infinite-volume QCD measure. Section d explicitly waives 'any issue of rigor in the construction of the various measures involved'. No argument is given that a truncated Gaussian or a delta distribution can actually arise as p_O for any local gauge-invariant observable in the QCD ensemble, nor that the required infinite-volume measures exist. As written, the examples invalidate the general moment-inference lemma for abstract random variables, but they do not by themselves show that the specific class of observables and assumptions used in Ref. [1] are inconsistent with Eq. (3). The author should either supply a realizability argument or explicitly frame the conclusion as shifting the burden of proof rather than as a full refutation.","section":"Sec. f, Eqs. (7)-(9)"}],"minor_comments":[{"comment":"The text says Eq. (3) holds for 'arbitrary non-negative integer l', but for l = 0 the statement is false because <O^0> = 1. The later text in Sec. g correctly restricts to l > 0; Eq. (3) and its surrounding discussion should be corrected accordingly.","section":"Sec. b, Eq. (3)"},{"comment":"The symbol P(m,A) is reused for the original non-negative quantity and for the rescaled quantity m^{-k0} P(m,A) chi_{S(O)}(A). Please introduce a distinct symbol, for example \tilde P(m,A), to avoid confusion in Eqs. (2)-(4).","section":"Sec. d"},{"comment":"The notation chi_{S_m(P)}(A) is not defined; the support set of the rescaled measure should be defined explicitly or written out in words.","section":"Sec. d"},{"comment":"The phrase 'for some non-negative integers l0, k' is ambiguous because k0 is introduced only later, and the relation k0 > k is stated after the definitions. It would be clearer to define k0 before Eq. (2) and state the inequality explicitly there.","section":"Sec. b, Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"The decisive point for the outcome is whether AFT's application after Eq. (92) actually uses l0 >= 2. If it uses l0 = 1, the manuscript's main objection collapses, since boundedness of O then gives the desired higher-moment suppression. Please ask the author to provide the exact passage and the value of l0 before further processing. Also, the abstract promises an additional note that is absent from the submitted text; the author should be required to either supply it or remove the claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe thing to know: this is a genuine, well-aimed shot at one step in Aoki-Fukaya-Taniguchi 2012. Giordano shows the inference from <O^{l0}> = O(m^{k0}) to <O^l> = O(m^{k0}) for all l is not a theorem. The truncated-Gaussian and delta-function density-of-states counterexamples are explicit, simple, and, as far as I know, new. That alone earns the paper a serious look.\n\nWhat is good: the comment does not just complain. It isolates Eqs. (23)-(25) of AFT, recasts the missing step in density-of-states language, and constructs distributions where one moment vanishes like m^{k0} while lower moments vanish more slowly. The point that positivity and shared support do not force the same-order suppression is correct, and the examples make it undeniable. The author also openly labels his measure construction heuristic, and the citation list covers both sides of the lattice controversy without obvious bias.\n\nWhere it is soft: the transfer to AFT's proof is not actually completed. Every counterexample involves l < l0. If AFT only use the lemma with l0 = 1, then for a bounded positive observable the conclusion for all l >= 1 follows trivially via O^l <= C^{l-1} O. Giordano asserts in Sec. G that the implication is crucially used after Eq. (92), but never states what l0 is there. Until he shows that AFT's application requires l0 > 1, the comment has a hole at exactly the load-bearing joint. Second, the abstract promises an additional note responding to Aoki-Fukaya's reply, and that note is absent from this version; that is a real completeness problem. Third, the infinite-volume measures and density-of-states functions are heuristic, so the counterexamples are mathematical analogies unless one argues they are realizable in QCD.\n\nNet assessment: the general lemma is dead; whether AFT's specific conclusion falls is still open. This is a legitimate comment that deserves peer review, but the referee should send it back for the author to verify the actual l0 in AFT's Eq. (92) neighborhood and to supply the promised note. If l0 turns out to be 1, the comment becomes a useful warning about a non-existent general proof, not a refutation of AFT's result.\n\nBest,","headline":"A real gap in AFT's moment-inference step, shown with clean new counterexamples, but the comment never confirms the actual l0 in their proof, so the refutation of their specific conclusion is incomplete.","tokens_in":6268,"tokens_out":5769,"would_cite":true,"duration_ms":52263,"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":"A crucial inference in a proof that chiral symmetry forces the Dirac spectral density and topological susceptibility to vanish at small quark mass is invalid: one power of a positive gauge-field observable vanishing like $m^{k_0}$ does…","keywords":["chiral symmetry restoration","Dirac operator spectral density","topological susceptibility","axial U(1) anomaly","finite-temperature QCD","density of states","lattice QCD","mass dependence of observables"],"falsifier":"On a lattice with two light quark flavors in the chirally symmetric phase, compute $p_O(x,m)=\\langle\\delta(x-O(A))\\rangle$ and the moments $\\langle O^l\\rangle$ for a positive local gluonic observable $O$ at several small masses $m$. If $p_O(x,m)$ has a peak whose position $x_0(m)$ vanishes as $m\\to 0$ while its width is comparable to $x_0(m)$, or more generally if the rescaled density $x_0^{-1}p_O(x/x_0,m)$ has a nonzero chiral limit, then the ratios $\\langle O^l\\rangle/\\langle O^{l_0}\\rangle$ will not stay of order one for different $l$, demonstrating that Eq. (1) does not imply Eq. (3).","tokens_in":5343,"feed_emoji":"⚛️","tokens_out":10185,"duration_ms":84559,"temperature":0.7,"pith_summary":"This comment targets a specific step in a proof that, in the chirally symmetric phase of two-flavor QCD, the Dirac spectral density at the origin and the topological susceptibility must vanish identically for sufficiently small nonzero quark mass $m$. The step claims that if one positive gauge-field observable satisfies $\\langle O(A)^{l_0}\\rangle=O(m^{k_0})$ as $m\\to 0$, then every power satisfies $\\langle O(A)^l\\rangle=O(m^{k_0})$ for all $l\\ge 1$. The comment shows, through density-of-states counterexamples, that this implication does not follow from positivity or shared support alone: a moment can vanish quickly because the value of $O$ on typical configurations shrinks, rather than because the contributing region of configuration space shrinks. If so, the original proof's conclusions $\\rho(0;m)=0$ and $\\chi_t=0$ require a missing argument. An appended note also refutes a reply's objection by showing that the assumptions invoked would force all local gluonic correlators to be independent of $m$ at small mass, which QCD does not exhibit.","feed_headline":"Key step fails in proof of vanishing QCD spectral density","feed_subtitle":"One vanishing power of a gauge-field observable does not force all powers to vanish, so the claimed zeros are unproven.","key_machinery":"The density-of-states function $p_O(x,m)=\\langle\\delta(x-O(A))\\rangle$ for a positive gauge-field observable $O(A)$ carries the argument. It separates two ways a moment can vanish: the support of $O$-values can shrink toward $x=0$ (as in $p_O=\\delta(x-x_0(m))$, where $\\langle O^l\\rangle=x_0(m)^l$), or the support of gauge-field configurations contributing to the expectation can shrink in measure while $O$ itself stays bounded, which is the original proof's hidden assumption. The counterexample distributions—a single delta moving to zero, a truncated Gaussian, and any density with a scaling form $p_O(x,m)=x_0(m)^{-1}\\pi_O(x/x_0(m),0)$—show that one moment's vanishing rate does not control the others. The key comparison is $\\langle O^l\\rangle/\\langle O^{l_0}\\rangle\\propto x_0^{l-l_0}$ or $\\sigma^{l-l_0}$, which blows up for $l<l_0$ whenever the relevant scale vanishes.","core_discovery":"The central claim, stated on the paper's own terms, is that Eq. (3) is not a necessary consequence of Eq. (1). Writing $\\langle O(A)^l\\rangle=\\int dx\\,x^l p_O(x,m)$ with $p_O(x,m)=\\langle\\delta(x-O(A))\\rangle$, the inference would require $p_O$ to have the special form $[1-\\alpha(m)]\\delta(x)+\\alpha(m)\\tilde p(x;m)$ with $\\alpha(m)=O(m^{k_0})$. But ordinary distributions invalidate this: if $p_O(x,m)=\\delta(x-x_0(m))$ with $x_0(m)\\to 0$, then $\\langle O^l\\rangle=x_0(m)^l$, so a single moment of order $l_0$ being $O(m^{k_0})$ does not make lower moments $O(m^{k_0})$ unless $l_0=l$. A truncated Gaussian with mean and width vanishing in the chiral limit behaves similarly; if the width-to-mean ratio has a nonzero or infinite chiral limit, ratios such as $\\langle O^l\\rangle/\\langle O^{l_0}\\rangle$ blow up for $l<l_0$. Thus the load-bearing inference in the original proof fails, and the claimed vanishing of $\\rho(0;m)$ and $\\chi_t$ for small nonzero $m$ is not established by that argument.","pith_inferences":["The counterexample construction transfers to any positive observable whose expectation value is controlled by a self-averaging value rather than by the size of the support; similar 'one moment controls all moments' steps in other spectral or thermodynamic arguments should be checked against the same density-of-states test.","A concrete lattice test suggested by the paper's own language: measure $p_O(x,m)$ for a simple positive gluonic quantity in the high-temperature phase; if the distribution shifts toward zero with a width comparable to its mean, the moments will scale with distinct powers of $m$, directly demonstrating the failure mode.","If the density-of-states distributions are not ordinary functions on the infinite-volume configuration space, the counterexamples may not transfer verbatim to QCD, but the burden is then on the original proof to justify the measure-theoretic setup it assumes; the logical gap itself is independent of that technical caveat.","The same type of argument could be tested in zero-dimensional or random-matrix models where the density of states is exactly known, providing a cheap check of whether the missing inference holds in simplified settings."],"forward_implications":["The specific proof in the original article no longer establishes $\\rho(0;m)=0$ or $\\chi_t=0$ for small nonzero $m$; these conclusions need a new argument that supplies the missing inference from one moment to all moments.","The claimed necessity of a phase transition at small nonzero quark mass in the high-temperature phase, drawn from $\\chi_t=0$, loses its proof-based support and remains an open numerical question.","The fate of the axial U(1) anomaly in the chiral limit is not settled by this proof route; the scalar-pseudoscalar sector analysis in the original article depends on the same unproven step.","Future proofs of the same conclusion must either impose a density-of-states structure (a fixed delta peak plus a smooth part with bounded moments) or otherwise control all moments, not just one.","The appended note adds a consistency constraint: any argument that both assumes $m^2$-analyticity of gluonic observables and uses the disputed implication would predict that all local gluonic correlators are mass-independent at small $m$, which contradicts observed QCD behavior."],"supporting_citations":[{"why":"Supplies the original proof step and conclusions that the comment argues are unjustified: its Eqs. (23)-(25) contain the inference from one power to all powers, used later for the vanishing claims.","marker":"[1]"},{"why":"Domain-wall fermion simulation cited as supporting the vanishing predictions for the spectral density toward the chiral limit.","marker":"[2]"},{"why":"Later lattice simulation cited among those supporting the vanishing predictions with controlled chiral and topological properties.","marker":"[4]"},{"why":"Early staggered-fermion study cited as reaching a different conclusion about the spectral density toward the chiral limit.","marker":"[6]"},{"why":"Staggered-fermion lattice study cited as disagreeing with the vanishing predictions.","marker":"[8]"},{"why":"Recent staggered-fermion simulation cited among results that reach different conclusions toward the chiral limit.","marker":"[10]"}],"fun_headline_variants":["Flawed step in QCD spectral density proof","No vanishing spectral density: moment inference fails","Chiral symmetry proof flawed: zero not forced","Spectral zero unproven: counterexample to key step","Key proof step invalid in Aoki-Fukaya-Taniguchi"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The counterexamples presume that the infinite-volume QCD expectation value has a well-defined measure on gauge-field configurations and that density-of-states functions behave like ordinary probability densities; if those measures do not exist or cannot be realized in QCD, the logical gap may remain real but the explicit counterexamples would not carry over.","fun_headline_variants_meta":{"raw":{"variants":["Flawed step in QCD spectral density proof","No vanishing spectral density: moment inference fails","Chiral symmetry proof flawed: zero not forced","Spectral zero unproven: counterexample to key step","Key proof step invalid in Aoki-Fukaya-Taniguchi"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000179,"raw_usage":{"total_tokens":1344,"prompt_tokens":1031,"completion_tokens":313,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":234}},"tokens_in":647,"tokens_out":313,"duration_ms":3674,"temperature":1.0,"reasoning_tokens":234,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:40:41.647140+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a lattice with two light quark flavors in the chirally symmetric phase, compute $p_O(x,m)=\\langle\\delta(x-O(A))\\rangle$ and the moments $\\langle O^l\\rangle$ for a positive local gluonic observable $O$ at several small masses $m$. If $p_O(x,m)$ has a peak whose position $x_0(m)$ vanishes as $m\\to 0$ while its width is comparable to $x_0(m)$, or more generally if the rescaled density $x_0^{-1}p_O(x/x_0,m)$ has a nonzero chiral limit, then the ratios $\\langle O^l\\rangle/\\langle O^{l_0}\\rangle$ will not stay of order one for different $l$, demonstrating that Eq. (1) does not imply Eq. (3).","supporting_citations":[],"review_version":2}