{"id":"aa73e29f-b8d3-49b8-aeab-8b93209add95","arxiv_id":"2608.13359","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For 0<p<1, the p-th moment of the critical 2D SHF mass on a small ball is bounded by the second moment raised to a negative power whenever the second moment diverges, uniformly in time, disorder, and radius.","lead":"This paper proves upper bounds on fractional moments of the mass that the critical 2D Stochastic Heat Flow and directed polymer partition functions assign to small balls, uniformly in time, disorder strength, and ball radius. A generalist reader might care because the bounds are governed entirely by the divergence of the second moment, giving a clean intermittency principle and a reusable coarse-graining tool for disordered systems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 3.3, Lemma 3.6, is the load-bearing point: it imports the uniform hypercontractivity estimate [CSZ25b, Thm 1.11] from an unpublished same-group preprint; if that theorem is not uniform along the coarse-graining scales, Proposition 2.3 and Theorem 1.1 lose their proof.","rationale":"The central claim (Theorem 1.1) is supported by a coherent two-stage strategy: a size-biased change of measure producing the suboptimal Proposition 2.3, followed by non-homogeneous coarse-graining in Proposition 2.4. I checked the internal estimates: Lemma 3.2 is correct; the variance decomposition (3.4) follows from the martingale property; the scale choices in §3.2 make Lemma 3.4 plausible via the second-moment bounds in Section 5; and the numerical threshold in Proposition 2.4 (K=4) is actually satisfied (the L1-ball count 41/300 plus a tail below 3.5e-3 is below e^{-1}). The serious soft spot is Lemma 3.6: the uniform third-moment control is imported from [CSZ25b, Thm 1.11], a preprint not proved here. If that theorem is not uniform in the exponentially growing coarse-graining scales, the diagonal and off-diagonal size-biased variance bounds cannot be combined, and Proposition 2.3 fails. The reader's verdict of CONDITIONAL is therefore appropriate: the architecture is sound conditional on the imported hypercontractivity estimate. Secondary issues (the one-line passage from polymer to SHF in §2.1, the citation of the withdrawn [Hua26] for the sharp lower bound) are real but addressable and do not affect the proof of Theorem 1.1 as directly. Recommendation: keep the conditional status, with the condition being a verified proof of Lemma 3.6 or an explicit statement of [CSZ25b, Thm 1.11] in the appendix.","tokens_in":28273,"tokens_out":25877,"duration_ms":219266,"concrete_test":"Locate [CSZ25b, Thm 1.11] and verify whether it states E[Z^{β,ω}_m(g)^3] ≤ C E[Z^{β,ω}_m(g)^2]^{3/2} for every probability mass function g, with C uniform in m,β and with no restriction on the smoothing of the initial condition. If the theorem is stated for the continuum SHF or with constants depending on the initial smoothing, then Lemma 3.6 has a genuine gap. A complementary check: independently prove that bound for g=f*q_a along the scales a=N_{i-1} and m=N_i-N_{i-1} of (3.6) using the chaos expansion; a non-uniform or failing bound would invalidate (3.10).","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.3, Lemma 3.6, is the point where the proof of Proposition 2.3 depends on a nontrivial input not established in this paper. Writing Z_i(f)=Z^{β,ω}_{N_{i-1},N_i}(f), the lemma needs ~Var_f[X_i(f)] ≤ E[Z_i(f)^3] ≤ C_3 E[Z_i(f)^2]^{3/2} uniformly over i=1..k, β∈(0,1), and the smoothed initial distributions f_{i-1}=f*q_{N_{i-1}} with f∈M^disc_1(R). The paper invokes [CSZ25b, Thm 1.11] for this 'hypercontractivity' property. That theorem is not stated or proved in the present manuscript, and [CSZ25b] is an unpublished preprint of the same group, arXiv:2504.06128. The load-bearing requirement is uniformity: the second moment is O(1) by Lemma 3.4 at every scale, but the third moment must be bounded by a universal constant times the 3/2 power of that second moment, with no dependence on N_{i-1} or the interval length. If the imported theorem holds only for fixed smoothing or for the continuum SHF, or with constants growing in the scale, then the size-biased variance bound (3.10) fails, Proposition 2.3 is unproven, and the coarse-graining argument in §2.3 has no basis. This is the least self-contained step in the proof.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper establishes uniform upper bounds for fractional moments p∈(0,1) of the mass assigned to small balls by the critical 2D Stochastic Heat Flow and its discrete analogue, 2D directed polymers, in the critical scaling regime. The central discrete statement (Theorem 1.5, with the equivalent half-moment Theorem 2.1) asserts that E[Z_N(U_R)^p] is bounded by a constant times E[Z_N(U_R)^2]^{c p(p-1)} uniformly in N, β∈(0,1), R≤√N, and the continuum version (Theorem 1.1) follows by scaling. The proof strategy is: (i) a size-bias change-of-measure lemma (Lemma 3.2) plus a proxy X built from disjoint disorder strips reduces the half-moment to variance and size-biased variance estimates (Proposition 2.3); (ii) a non-homogeneous coarse-graining procedure (Proposition 2.4) boosts the suboptimal estimate to the final decay by choosing scales adapted to R, N, and ϑ; (iii) Section 5 provides sharp second-moment estimates for the polymer and SHF (Propositions 1.6 and 1.2), derived from a soft-tilt renewal identity (Lemma 5.1).","tokens_in":28598,"tokens_out":15379,"duration_ms":137513,"significance":"If the proof is completed, this is a substantial contribution to the study of the critical 2D Stochastic Heat Flow and directed polymers. It provides the first bounds on fractional moments of small-ball masses that are uniform in all parameters, identifies the divergence of the second moment as the mechanism forcing fractional moments to vanish, and supplies sharp uniform second-moment estimates (Proposition 1.6 and Lemma 5.1) of independent interest. The new ingredients — the size-bias proxy with non-homogeneous scales and the soft-tilt renewal identity — are natural and reusable. The main obstacles are not conceptual but completeness: two central estimates are imported without proof from unpublished same-group preprints, and the passage from the discrete theorem to the continuum theorem is only sketched. With those points addressed, the paper would be an important advance.","major_comments":[{"comment":"The sentence 'The proof follows from Theorem 1.5 by taking the limit N→∞ and using Theorem 1.4' is not a proof. To pass from the discrete estimate (1.13)-(1.14) to (1.2)-(1.3) one must justify convergence of E[(Z_{[Nt]}(U_{R_N}))^p] to E[Z_t^ϑ(U_r)^p] under the distributional convergence of Theorem 1.4; this requires either a uniform integrability argument or a Fatou-type lower-semicontinuity bound, since z↦z^p is unbounded above. The authors must also specify the scaling R_N≈√N r and verify that ϑ([Nt],β_N) converges to ϑ+log t and that log(N/(1+R_N^2)) is comparable to log(1+t/r^2), so that the right-hand side of (1.14) converges to the right-hand side of (1.3). As written, Theorem 1.1 is not established.","section":"Section 2.1, proof of Theorem 1.1"},{"comment":"The bound ~Var_f[X_i(f)] ≤ E[Z_i(f)^3] ≤ C_3 E[Z_i(f)^2]^{3/2} is the only estimate controlling the diagonal size-bias variance, and it is imported from [CSZ25b, Thm. 1.11] without stating that theorem or its hypotheses. The required uniformity is over i=1,...,k, β∈(0,1), and the smoothed initial laws f*q_{N_{i-1}} with f∈M^disc_1(R); this is precisely the scale-uniformity that Proposition 2.3 needs. Since [CSZ25b] is an unpublished same-group preprint and the theorem is not reproduced, this step is not verifiable from the manuscript. The authors should either state the theorem with all hypotheses or prove the needed uniform hypercontractivity in a self-contained appendix.","section":"Section 3.3, proof of Lemma 3.6"},{"comment":"Two further load-bearing inputs are imported without proof from the unpublished preprint [BCT25]. In the proof of Lemma 3.7, the inequality q(f)_{2a_2-b_1}(y_1) - q_{2a_2}(f,f) ≤ C b_1/(a_2)^2 is used directly as '[BCT25, Eq. (6.4)]'; this bound produces the crucial factor N_i/N_{j-1} in the off-diagonal covariance estimate. In Section 2.3.2, the supercritical case of Theorem 2.1 concludes with '[BCT25, Thm. 2.2]'. Neither statement is reproduced, and [BCT25] is not yet refereed. These estimates should be stated explicitly and proved here, or replaced by self-contained derivations.","section":"Section 3.3, proof of Lemma 3.7, and Section 2.3.2"}],"minor_comments":[{"comment":"The text says a result was 'announced in [Hua26]', but the reference list describes [Hua26] as 'withdrawn pending major revision'. The status of this preprint should be qualified in the text, and the result should not be presented as an established announcement.","section":"Section 1.2 and References"},{"comment":"The phrase 'Theorem Theorem 1.5' contains a duplicated word; also, the sentence 'The proof follows from Theorem Theorem 1.5...' should be expanded into a short argument, as discussed in the first major comment.","section":"Section 2.1, proof of Theorem 1.1"},{"comment":"The equality case Z≡1 makes the right-hand side of (3.1) an indeterminate expression 0/0; the remark should explain that this case is obtained by a limiting or trivial interpretation.","section":"Section 3.1, Remark 3.3"},{"comment":"The proof uses two unshown numerical evaluations: 'for K=4 the sum of the two terms is smaller than e^{-1}' and 'the last inequality follows from a numerical evaluation' for ∑_y Q_0(y) ≤ 20. These are explicit sums over the l1 lattice; the authors should provide a displayed calculation or a simple closed-form bound.","section":"Section 4, proof of Proposition 2.4"},{"comment":"The monotonicity in β used to compare Z^{β,ω} with Z^{β_c,ω} is cited to [Zyg24, Prop. 3.4] without statement; since this is a separate imported result, it would help to state the precise monotonicity property being used.","section":"Section 2.3.2"}],"recommendation":"major_revision","confidential_remarks":"This is a promising and substantial paper, but several central estimates are imported from unpublished preprints of the same group ([CSZ25b], [BCT25], and also [CSZ25a] in Section 5.1). The editor may wish to verify whether the relevant theorems have appeared in refereed venues; if not, the authors should be required to state them fully or prove them. The withdrawal status of [Hua26] should also be handled carefully in the revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zygouras and coauthors have a real result. The uniform-in-(N,beta,R) and (t,theta,r) fractional moment bounds are new, and the paper earns them: the inhomogeneous coarse-graining and the soft-tilt renewal identity are genuine technical tools, and the sharp second moment estimates in Section 5 are useful on their own. The architecture is coherent: size-biased change of measure reduces to variance control, coarse-graining turns a sub-optimal half-moment bound into the stated decay, and the second moment estimates are handled in detail. This is not a re-coordinatization of known results.\n\nThe main soft spot is exactly the one the stress test names. Lemma 3.6 imports hypercontractivity from [CSZ25b, Thm. 1.11], and that theorem is from the same group's unpublished preprint. The proof needs uniformity over the smoothed initial laws f*q_{N_{i-1}} and over the coarse-graining scales, with constants independent of interval length. If the imported theorem does not provide that uniformity, the size-biased variance bound and Proposition 2.3 fail. The paper does not state the theorem or quote its precise form, so a referee must verify this point directly. I would not call it a flaw in the architecture—everything else hangs together—but it is load-bearing and not self-contained.\n\nSeveral smaller issues: the passage from Theorem 1.5 to Theorem 1.1 is one sentence, and for a paper whose headline is the SHF result that limit deserves at least a paragraph. The numerical evaluations in Proposition 2.4 are asserted without tables; they look harmless, but the values should be included. And [Hua26] is listed in the bibliography as withdrawn yet cited in the text as if it had established (1.6). That should be fixed before publication; the paper can still discuss the announced result, but should not present it as settled.\n\nThe proof is not circular: the fractional moment bound is not assumed, and the second moment estimates are proven independently. The central mathematical claim is well supported. This paper is for probabilists working on 2D polymers, the SHF, and intermittency. It deserves serious refereeing, conditional on the imported hypercontractivity being checked. I would send it out and would want the referee to confirm the uniformity in Lemma 3.6 before acceptance.","headline":"Genuinely new uniform fractional moment bounds for the critical 2D SHF and polymers, with a sound two-stage proof; the main risk is a load-bearing import from an unpublished same-group preprint.","tokens_in":29147,"tokens_out":2007,"would_cite":true,"duration_ms":20229,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B44","60K35","82D60"],"pacs":[],"model":"deepseek-v4-flash","headline":"In the critical 2D stochastic heat flow, the divergence of the second moment forces every fractional moment of a small ball's mass to zero.","keywords":["Stochastic Heat Flow","directed polymer","fractional moments","second moment estimates","Volterra function","coarse-graining","change of measure","intermittency"],"falsifier":"One concrete check is a lattice simulation of the 2D directed polymer at fixed large $N$ with $R=0$ and disorder tuned so $e^{\\vartheta(N,\\beta)}$ is large: the paper predicts $E[Z^{\\beta,\\omega}_N(\\delta_0)^{1/2}]\\le C\\,V(e^{\\vartheta})^{-c}$, i.e. a half-moment that is numerically tiny. If the sample-averaged half-moment remains of order one while the sample second moment is visibly diverging, Theorem 1.5 is false. A more direct mathematical check is to test whether the third-moment control of [CSZ25b, Theorem 1.11] is uniform in the initial law $f*q_a$ for $a$ running over the specific coarse-graining scales $N_i-N_{i-1}$; a counterexample there would break Lemma 3.6.","tokens_in":28056,"feed_emoji":"📉","tokens_out":16211,"duration_ms":135077,"temperature":0.7,"pith_summary":"The paper establishes that the fractional moments of the random mass assigned to small balls by the critical 2D Stochastic Heat Flow, and by the 2D directed polymer model of which the flow is the scaling limit, are governed by the second moment alone: for every $p\\in(0,1)$, $E[Z^\\vartheta_t(U_r)^p]\\le C E[Z^\\vartheta_t(U_r)^2]^{c p(p-1)}$, uniformly in time $t$, disorder strength $\\vartheta$ and ball radius $r\\le\\sqrt{t}$, with $U_r$ the uniform distribution on a ball of radius $r$. Because $p-1<0$, the bound goes to zero exactly when the second moment diverges, so the small-ball mass becomes vanishingly small in probability in every regime where the second moment blows up. This matters because non-integer moments of these objects resist the standard tools that work for integer moments, and the paper reduces the fractional problem to a second-moment estimate plus a new multiscale iteration. It also supplies sharp uniform second-moment bounds, with the Volterra function as the control parameter, that are of independent interest.","feed_headline":"Second-moment divergence forces all fractional moments to zero","feed_subtitle":"Uniform bounds show polymer and continuum small-ball masses vanish exactly where their second moment blows up.","key_machinery":"The carrying mechanism is a size-biased change of measure. For a mean-one random variable $Z$, Lemma 3.2 proves that $E[\\sqrt Z]\\le \\sqrt{2\\Sigma^2/(2\\Sigma^2+\\Delta^2)}$, where $\\Delta$ is the shift in the mean of an arbitrary proxy $X$ when the law is tilted by $Z$, and $\\Sigma^2$ is the sum of its variances under the original and tilted laws. The proxy is a sum of centered partition functions on dyadic scales $N_i=N^{1-\\alpha_0 2^{-i}}$, chosen so each block contributes order-one variance and the size-biased mean of the proxy equals its variance; a lower bound on that variance and an upper bound on the size-biased variance yield the sub-optimal decay. A non-homogeneous coarse-graining with growing blocks $M_i>M_{i-1}$ amplifies this one-block contraction along an exponential ladder of scales, which is the paper's main methodological step. The second-moment estimates are carried by a soft-tilt renewal identity for the polymer variance, whose asymptotic evaluation at the critical tilt produces the Volterra function $V(T)=\\int_0^\\infty T^s/\\Gamma(s+1)\\,ds$, the quantity that interpolates between $e^T$ and $1/\\log(1/T)$ and controls the divergent regimes.","core_discovery":"The paper's central claim, Theorem 1.1 for the continuum Stochastic Heat Flow and Theorem 1.5 for the directed polymer model, is a uniform reduction of fractional moments to the second moment. The precise inequality is $E[Z^\\vartheta_t(U_r)^p]\\le C E[Z^\\vartheta_t(U_r)^2]^{c p(p-1)}$, and the discrete analogue holds for $E[Z^{\\beta,\\omega}_N(U_R)^p]$ uniformly in $N$, $\\beta\\in(0,1)$, and $R\\le\\sqrt{N}$. Since the exponent $c p(p-1)$ is negative, the right side tends to zero whenever the second moment diverges; the paper's conclusion is that the divergence of the second moment is the single mechanism behind the vanishing of small-ball masses, a signature of strong intermittency. The proof works through the polymer model: a size-biased change-of-measure estimate (Proposition 2.3) gives a quantitative but weak decay of the half-moment on individual blocks, and a new non-homogeneous coarse-graining (Proposition 2.4) with growing scales iterates that decay into the final bound. The continuum statement follows by taking the scaling limit of polymer partition functions, and the sharp second-moment estimates (Propositions 1.2 and 1.6) identify the Volterra function as the precise quantifier of the divergent regime.","pith_inferences":["If the paper's Conjecture 1.3 is correct, the same dominance of the second moment would extend to all real $p$, including negative moments; the paper proves only $p\\in(0,1)$, but the approximate log-normality of small-ball masses makes negative moments a natural next testable target.","The non-homogeneous coarse-graining with growing blocks, whose scale ratio is tuned to the disorder parameter, is not model-specific: I would expect it to transfer to other marginally relevant disordered systems where integer or second moments are computable but fractional moments are not.","A direct consequence that could be probed numerically is that the half-moment of the point-to-plane polymer partition function decays on the scale $\\exp(-c e^{\\vartheta})$ whenever the second moment diverges; simulations across the critical window at moderate $N$ would check whether the uniform constants in Theorem 1.5 are of the right order."],"forward_implications":["Whenever the second moment of the normalized small-ball mass diverges---for example when $t e^{\\vartheta}\\to\\infty$ or when $r/\\sqrt{t}$ shrinks fast---every fractional moment of order $p\\in(0,1)$ tends to zero, and the mass converges to $0$ in probability.","In the discrete polymer model the same conclusion holds uniformly in $N$, $\\beta$ and starting radius $R\\le\\sqrt{N}$, including the point-to-plane case $R=0$.","The sharp two-sided variance estimates extend across subcritical, critical and supercritical disorder, identifying the Volterra function $V(e^{\\vartheta(N,\\beta)})$ as the parameter that controls fluctuations in all regimes.","Through the scaling limit, the polymer-level bound passes to the continuum Stochastic Heat Flow, so the discrete proof is the engine behind Theorem 1.1.","In the multifractal formalism, the expected $p$-th power sum over $r$-balls behaves like $r^{2p-2}(\\log(1/r))^{c p(p-1)+o(1)}$, giving logarithmic corrections to the Euclidean counting exponent $2p-2$."],"supporting_citations":[{"why":"Supplies the hypercontractivity theorem (Theorem 1.11) that bounds third moments of polymer partition functions with diffused initial laws by powers of their second moments; this is the key input for the diagonal size-biased variance estimate Lemma 3.6.","marker":"[CSZ25b]"},{"why":"Provides the change-of-measure framework, an earlier constant-scale coarse-graining, and the supercritical half-moment bound (Theorem 2.2) used to close the case $\\vartheta(N,\\beta)>0$.","marker":"[BCT25]"},{"why":"Constructs the Stochastic Heat Flow as the scaling limit of critical polymer partition functions (Theorem 1.4), which transfers Theorem 1.5 from the discrete polymer to the continuum Theorem 1.1.","marker":"[CSZ23]"},{"why":"Gives the renewal-theoretic identity for the second moment of polymer partition functions and the log-asymptotics of two-dimensional return probabilities used in Lemma 5.1 and in the sharp variance bounds.","marker":"[CSZ19b]"},{"why":"Supplies the sharp small-$\\beta$ asymptotic of the Lyapunov exponent $F_2(\\beta)$ used in Lemma 5.1 to prove the second-moment estimates outside the critical window.","marker":"[BN26]"},{"why":"Provides the monotonicity of fractional moments in $\\beta$ used to reduce the supercritical regime to the critical value $\\vartheta=0$.","marker":"[Zyg24]"}],"fun_headline_variants":["Fractional moments vanish iff second moment diverges","Second-moment blow-up sets all fractional moments to zero","Uniform law: small-ball masses die with second moment","Uniform criterion: second moment divergence controls fractional moments"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is an imported hypercontractivity bound, [CSZ25b, Theorem 1.11], used in Lemma 3.6: it says the third moment of a polymer partition function with a diffused starting distribution is controlled by its second moment, and the proof needs that control to hold with one uniform constant along every growing coarse-graining scale; if the bound fails or the constant degrades, the key sub-optimal estimate and the whole iteration collapse.","fun_headline_variants_meta":{"raw":{"variants":["Fractional moments vanish iff second moment diverges","Second-moment blow-up sets all fractional moments to zero","Uniform law: small-ball masses die with second moment","Uniform criterion: second moment divergence controls fractional moments"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00067,"raw_usage":{"total_tokens":3044,"prompt_tokens":927,"completion_tokens":2117,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":543,"completion_tokens_details":{"reasoning_tokens":2055}},"tokens_in":543,"tokens_out":2117,"duration_ms":16469,"temperature":1.0,"reasoning_tokens":2055,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:55:28.741244+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete check is a lattice simulation of the 2D directed polymer at fixed large $N$ with $R=0$ and disorder tuned so $e^{\\vartheta(N,\\beta)}$ is large: the paper predicts $E[Z^{\\beta,\\omega}_N(\\delta_0)^{1/2}]\\le C\\,V(e^{\\vartheta})^{-c}$, i.e. a half-moment that is numerically tiny. If the sample-averaged half-moment remains of order one while the sample second moment is visibly diverging, Theorem 1.5 is false. A more direct mathematical check is to test whether the third-moment control of [CSZ25b, Theorem 1.11] is uniform in the initial law $f*q_a$ for $a$ running over the specific coarse-graining scales $N_i-N_{i-1}$; a counterexample there would break Lemma 3.6.","supporting_citations":[],"review_version":1}