{"id":"0fa80917-03cd-4945-bc3a-36405b676f31","arxiv_id":"1908.07120","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Critical continuum random polymer measures M_r are constructed on diamond fractals, and intersections of two independent paths are shown to have Hausdorff dimension zero with log-Hausdorff exponent 1.","lead":"This paper constructs a family of continuum random polymer measures on a two-dimensional diamond fractal and studies how often two random paths intersect. The disordered measure forces pairs of paths to have uncountably many intersections of Hausdorff dimension zero, a strong localization effect.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central construction of M_r depends entirely on the unproved array theorem from companion preprint [11]; a gap there would invalidate Theorem 2.11 and the downstream results.","rationale":"The reader's weakest-assumption analysis correctly identifies the dependence on Theorem 5.2 from the companion preprint. My reading of the manuscript confirms that this is the decisive external input: Theorem 2.11 constructs M_r from the arrays supplied by Theorem 5.2, and the uniqueness proof reduces directly to uniqueness of those arrays. I did not find a more serious internal flaw in the proofs of the properties that follow from Theorem 2.11; the martingale arguments, the measure extensions, and the population-model computations are coherent as far as they go. The concern is not that the paper is internally inconsistent, but that its central existence/uniqueness claim is unverifiable from the text alone. Since the companion preprint is available and the paper explicitly flags the dependence, the appropriate disposition is the same conditional verdict already given by the reader: the construction should be accepted only after Theorem 5.2 is verified or supplied. Hence no further change to the verdict is needed.","tokens_in":42874,"tokens_out":11472,"duration_ms":127548,"concrete_test":"Obtain arXiv:1908.06555 and re-derive Theorem 3.12 line by line. Specifically verify that the limiting arrays {W_e^(k)} are well-defined as almost-sure or L^2 limits of the finite-n construction, are nonnegative, have the stated mean, variance R(r−k), and centered moments R^(m)(r−k), and satisfy the recursion W_e^(k) = (1/b)Σ_i Π_j W_{e×(i,j)}^{(k+1)} almost surely for every k and e—not merely in distribution. Then re-run the proof of Theorem 2.11 with this derivation substituted for the citation. If the companion proof requires additional assumptions or contains a gap, the manuscript should include a self-contained proof of Theorem 5.2 before the construction of M_r is accepted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5.1 states: 'The proof of Theorem 2.11 below relies on Theorem 5.2, which was proven in [11].' Theorem 5.2 is not proved or derived in this paper. It supplies the exact objects from which M_r is built: for each r, an infinite sequence of edge-labeled i.i.d. arrays W_e^(k) with mean one, variance R(r−k), prescribed centered moments, and the a.s. recursion W_e^(k) = (1/b)Σ_i Π_j W_{e×(i,j)}^{(k+1)}. This is a nontrivial distributional fixed-point statement. In the proof of Theorem 2.11, M_r(A) is defined as a product of W_e^(N) over the edges of a coarse-grained path, and properties (I)–(IV) are verified by direct appeal to Theorem 5.2. The uniqueness half of Theorem 2.11 is logically equivalent to the uniqueness half of Theorem 5.2: any family satisfying (I)–(IV) produces arrays of total masses of the component measures, and those arrays must have the unique law asserted in [11]. Thus every later theorem—singularity to μ, non-atomicity, intersection-set log-Hausdorff exponent, the ϑ_{M_r} correlation asymptotics, and the Hilbert–Schmidt/trace-class dichotomy—inherits its validity from an external preprint that is not included. This is not an internal inconsistency, but it is the single most load-bearing unverified premise. If Theorem 3.12 of [11] is false, or if it requires hypotheses beyond the stated moment conditions, the central claim of the present paper has no support within this manuscript.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a one-parameter family of random measures (M_r)_{r∈R} on the space of directed paths of a diamond fractal of Hausdorff dimension two, intended as continuum limits of critical weak-disorder polymer models. The construction is based on an external theorem on edge-labeled arrays; from these arrays the paper defines M_r by product weights on cylinder sets and verifies expectation μ, correlation measure υ_r, moment formulas, and a recursive distributional identity. The main probabilistic results are: M_r is a.s. singular to μ and non-atomic; under M_r×M_r, pairs of paths have intersection-time sets that are a.s. uncountable of Hausdorff dimension zero, with log-Hausdorff exponent exactly one on a positive-measure event; the associated spatial intersection measure ϑ_Mr on the fractal has Hausdorff dimension two and finite logarithmic energy only for exponent λ>9; and the intersection-time kernel defines a Hilbert-Schmidt, non-trace-class operator on L^2(Γ,M_r). The paper also states a weak-disorder continuum limit theorem whose proof reduces to an external array convergence theorem.","tokens_in":43228,"tokens_out":13415,"duration_ms":135827,"significance":"If the external premises hold, the paper gives a rare explicit continuum model for marginally relevant disorder in a hierarchical setting, with exact renormalization symmetry and precise intersection-set dimension. The self-contained parts—correlation measure construction, martingale analysis, population-model estimates, energy bounds, and the operator factorization—are detailed and mostly convincing. The main caveat is that the existence and uniqueness of the central random measures is not proved in this manuscript; it is imported from the author's companion preprint. The paper is therefore best read as a conditional construction whose value depends on the companion result being made available and verified.","major_comments":[{"comment":"The proof of Theorem 2.11 begins by stating that it relies on Theorem 5.2, which was proven in [11]. Theorem 5.2 supplies the i.i.d. edge-labeled arrays W_e^{(k)} with mean one, variance R(r-k), prescribed centered moments, and the a.s. recursion W_e^{(k)}=(1/b)Σ_i Π_j W_{e×(i,j)}^{(k+1)}. The measure M_r is then defined as a product of these W's over coarse-grained edges, and the uniqueness argument identifies any competing family's total masses with the unique array law from Theorem 5.2. Thus Theorem 2.11 is logically equivalent to the imported array theorem, and Theorem 2.22 is likewise reduced to Theorem 5.5 from [11]. Because [11] is a preprint not included in the submission, the central claim is not self-contained. The paper should either prove Theorem 5.2 and Theorem 5.5 here, or give a precise published reference that supplies them.","section":"§5.1 and §5.3"},{"comment":"The uniqueness argument assumes that any family (M_r) satisfying properties (I)-(IV) admits a consistent edge-indexed family (M^e_{r-k}) as in Corollary 2.14. Corollary 2.14 is stated without proof, and it is not immediate from property (IV) alone because that property is only an equality in distribution for the single measure, not a construction of an a.s. coupled collection indexed by all e∈∪E_k. Since the uniqueness half of Theorem 2.11 depends on this step, a proof of Corollary 2.14, or another argument establishing the array recursion a.s., is required.","section":"§5.1, uniqueness half of Theorem 2.11"}],"minor_comments":[{"comment":"The sentence 'ϱ_n(p)=ϱ_n(p) for every p∈Γ_n' is a tautology; the intended definition of the averaged measure on Γ should be stated with a fresh symbol or with explicit reference to M^ω_{β,n}.","section":"Definition 2.21"},{"comment":"Part (v) of Proposition 2.16 is stated as a result of this paper, but its proof is deferred to [12]; either include the proof here or label the statement as imported from the companion preprint.","section":"Proposition 2.16(v)"},{"comment":"The probability expression '1/b(R(r-n-1))^b/(1+R(r-n))' is missing parentheses; it should be written as (1/b)(R(r-n-1))^b/(1+R(r-n)) to avoid ambiguity.","section":"§6.1"},{"comment":"The notation in part (iii) switches between E_{υr} and E_{~υr}; the proof should use one probability measure consistently, since ~υr is not defined in the statement.","section":"Lemma 6.4"},{"comment":"The text refers to an illustration of the first few diamond graphs, but no figure appears in the manuscript; either include the figure or delete the reference.","section":"Section 1"}],"recommendation":"major_revision","confidential_remarks":"The central construction depends heavily on two unpublished companion preprints by the same author, [11] and [12]; the present manuscript is not self-contained. I recommend asking the author to include proofs of the imported array theorems or to postpone submission until [11] is published or otherwise made available to referees. There is no evidence of data fitting or normalization-forcing; the internal proofs appear coherent once the external premises are granted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nHere is my take. The genuinely new content is the family of critical continuum random polymer measures and, more importantly, the analysis of their intersection-time sets: Lemma 2.26 and Theorem 2.27 give the log-Hausdorff exponent exactly 1, and the proof goes through the correlation measure, martingale convergence, and an energy bound (Prop 6.7) that looks correct. I also found the ϑ_Mr correlation asymptotics in Theorem 2.34 and the Hilbert–Schmidt/trace-class decomposition in Theorem 2.41 careful and plausible. Section 4, where the correlation measure is built from first principles, is self-contained and rigorous.\n\nThe soft spot is the one the reader flagged: Theorem 2.11, the existence and uniqueness of M_r, is not proved here. It relies on Theorem 5.2 (Theorem 3.12 of [11]), which supplies the edge-labeled arrays with exact recursion and prescribed moments. The paper says so explicitly, so there is no pretense, but the uniqueness half of 2.11 is basically equivalent to the uniqueness half of 5.2. If the array theorem has a gap, the construction of M_r and every later theorem has no support inside this manuscript. The variance function R(r) is likewise imported from [9]. Neither proof is reproduced.\n\nThere are minor issues: Proposition 2.16(v)'s proof is in [12], and in the proof of Theorem 2.22, 'By Theorem 2.22' should presumably be 'By Theorem 5.5' — a typo, but confusing.\n\nShould a referee engage? Yes. The self-contained parts are solid and the intersection-time analysis is a real contribution. But the referee must have access to [11] and [9], or the author should be asked to include the array construction (or at least a detailed proof sketch) in a revised version. Without that, the paper is a conditional extension of the companion work, not an independent object.\n\nFor whom: specialists in directed polymers on hierarchical lattices and fractal disordered systems. It would be a good reading-group paper if [11] is on the table. I would not desk-reject it.\n\nBest.","headline":"The critical CRPM construction and the log-Hausdorff intersection analysis are genuinely new and mostly solid, but Theorem 2.11 is imported wholesale from the companion paper [11], so the manuscript is a conditional extension rather than a self-contained proof.","tokens_in":43709,"tokens_out":2691,"would_cite":false,"duration_ms":28925,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K35","28A80","60G57","28A78","82D60"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs a unique one-parameter family of continuum random polymer measures on a two-dimensional diamond fractal and shows that, under their realizations, two independent paths intersect in uncountable sets of Hausdorff…","keywords":["directed polymers","continuum random polymer","diamond fractal","critical disorder","weak-disorder scaling","Hausdorff dimension","log-Hausdorff exponent","hierarchical lattice"],"falsifier":"Compute the finite-$n$ pair correlation $\\mathbb{E}[M_{r,n}(p)M_{r,n}(q)]$ for two cylinder paths in the discrete model of Theorem 2.22; Lemma 2.7 predicts it equals $|\\Gamma_n|^{-2}(1+R(r-n))^{\\xi_n(p,q)}$. A measured deviation from this formula for large $n$, or a direct construction of two different array laws satisfying Theorem 5.2 for the same $r$, would disprove the uniqueness or existence claim.","tokens_in":42677,"feed_emoji":"💠","tokens_out":8104,"duration_ms":84855,"temperature":0.7,"pith_summary":"This paper aims to construct the continuum random polymer measures that arise as the weak-disorder scaling limit of directed polymers on diamond hierarchical graphs exactly at marginal relevance, where the graph has Hausdorff dimension two. It claims there is a unique one-parameter family of random measures on the space of directed paths with prescribed expectation, correlations, and a recursive self-similarity identity. If the construction is right, the critical measures are qualitatively different from their subcritical counterparts: although two paths drawn from the uniform measure almost surely cross only finitely often, paths drawn from a single realization of the disordered measure almost surely have positively weighted intersection sets that are uncountable yet of Hausdorff dimension zero. The paper further proves these zero-dimensional sets have logarithmic Hausdorff exponent exactly one. A sympathetic reader would care because this gives a concrete, exactly solvable model for the marginally relevant disorder regime, where a Gaussian multiplicative chaos representation is impossible.","feed_headline":"At criticality, paths meet on fractal dust","feed_subtitle":"Unique random polymer measures force independent paths to share uncountable, dimension-zero sets.","key_machinery":"The load-bearing objects are the edge-labeled random arrays $\\{W_e^{(k)}\\}_{e\\in E_k}$, i.i.d. within each generation, with mean one, variance $R(r-k)$, and the exact recursion $W_e^{(k)}=\\frac{1}{b}\\sum_i\\prod_j W_{e\\times(i,j)}^{(k+1)}$. The variance function $R(r)$ satisfies $R(r+1)=\\frac{1}{b}[(1+R(r))^b-1]$ with prescribed asymptotics, and it fixes the total-mass fluctuations while entering the cylinder formula for the correlation measure $\\upsilon_r(p\\times q)=|\\Gamma_n|^{-2}(1+R(r-n))^{\\xi_n(p,q)}$. The martingale $\\varphi_n^{(r,t)}$ links the correlation measures at different parameters and produces the intersection-time kernel $T(p,q)=\\lim_{n\\to\\infty}\\frac{\\kappa^2}{n^2}\\xi_n(p,q)$. A generation-inhomogeneous population model for shared edges $\\xi_n$ and surviving lineage count $\\tilde{\\xi}_n$ translates intersection-set sizes into a critical branching process, yielding the logarithmic energy bound and the log-Hausdorff exponent one.","core_discovery":"The central discovery is the existence and uniqueness of the random measures $M_r$ on the path space $\\Gamma$ of the diamond fractal of Hausdorff dimension two, satisfying $\\mathbb{E}[M_r]=\\mu$, $\\mathbb{E}[M_r\\times M_r]=\\upsilon_r$, finite centered moments of total mass with variance $R(r)$, and the recursive distributional identity $M_{r+1}\\overset{d}{=}\\frac{1}{b}\\sum_{i=1}^{b}\\prod_{j=1}^{b} M_r^{(i,j)}$. The construction starts from edge-labeled random arrays $W_e^{(k)}$ with mean one, variance $R(r-k)$, and the exact recursion $W_e^{(k)}=\\frac{1}{b}\\sum_{i=1}^{b}\\prod_{j=1}^{b} W_{e\\times(i,j)}^{(k+1)}$; the measure of a cylinder path is a normalized product of array entries. The paper then shows that for a.e. realization of $M_r$, the product $M_r\\times M_r$ assigns positive weight to pairs of paths whose intersection-time set is uncountable but of Hausdorff dimension zero, with log-Hausdorff exponent exactly one, and that the law of $M_r$ cannot be represented as a subcritical Gaussian multiplicative chaos because the formal coupling strength would have to be infinite.","pith_inferences":["If the construction is correct, the exact log-Hausdorff exponent one is a quantitative prediction: Monte Carlo sampling of finite-$n$ critical polymers on diamond graphs should show that surviving shared-edge lineages grow linearly in $n$, so the minimal cover count of the intersection set grows like $n$, not $n^2$.","The same array-recursion method may generalize to other exactly renormalizable hierarchical graphs of Hausdorff dimension two, where the same critical population transition would force path intersections onto dimension-zero sets.","The Hilbert-Schmidt but not trace-class dichotomy for $T_{M_r}$ suggests that the natural Gaussian field on $(\\Gamma,M_r)$ has rougher sample paths than in the subcritical GMC picture; this could be tested through the a.s. modulus of continuity of the conditional field constructed from $M_R$ for $R<r$.","The critical CRPMs here provide a hierarchical toy model for the critical (2+1)-dimensional stochastic heat equation, where no exact GMC construction is available; the intersections of independent paths under $M_r$ may be the analogue of the collision structure expected in that setting."],"forward_implications":["For every real $r$ there is exactly one law of random measures $M_r$ satisfying the four axioms of Theorem 2.11, realized by the product construction from edge-labeled arrays.","For a.e. realization of $M_r$, the product $M_r\\times M_r$ gives positive weight to path pairs whose intersection-time set is either finite or uncountable of Hausdorff dimension zero; in the second case the log-Hausdorff exponent is exactly one.","$M_r$ is a.s. non-atomic and mutually singular to the uniform path measure, has dense support, converges to the uniform measure as $r\\to-\\infty$, and its total mass converges in probability to zero as $r\\to\\infty$.","The continuum measures are the $n\\to\\infty$ weak-disorder limits of finite polymer partition-function measures at the critical scaling $\\beta_{n,r}=\\kappa n^{-1/2}-\\tau\\kappa^2/(2n)+\\kappa\\eta\\log n/n^{3/2}+\\kappa r/n^{3/2}+o(n^{-3/2})$.","The induced spatial intersection measure $\\vartheta_{M_r}$ has expectation $R'(r)\\nu$, is a.s. of Hausdorff dimension two, has logarithmic energy finite exactly for $\\lambda>9$, and the intersection kernel defines a Hilbert-Schmidt but not trace-class operator on $L^2(\\Gamma,M_r)$."],"supporting_citations":[{"why":"Supplies Theorem 5.2 (existence and uniqueness of the edge-labeled arrays) and Theorem 5.5 (convergence of discrete arrays), on which Theorem 2.11 and Theorem 2.22 directly rest.","marker":"[11]"},{"why":"Supplies Lemma 2.5, the unique variance function $R(r)$ with its recursion and asymptotics, which underpins all variance and correlation formulas.","marker":"[9]"},{"why":"Supplies the diamond hierarchical lattice as a compact metric path space, the uniform measure $\\mu$, and the subcritical CRPM construction that the critical case extends.","marker":"[10]"},{"why":"Provides the proof that the total mass $M_r(\\Gamma)$ converges to zero in probability as $r\\to\\infty$ and the conditional GMC representation used in Proposition 2.16.","marker":"[12]"},{"why":"Gives the critical-window scaling for the (2+1)-dimensional directed polymer that motivates the form of $\\beta_{n,r}$ in Definition 2.19 and situates the hierarchical model among marginally relevant disordered systems.","marker":"[8]"},{"why":"Defines generalized Hausdorff measures used to quantify the zero-dimensional intersection sets and to state the log-Hausdorff exponent results.","marker":"[13]"}],"fun_headline_variants":["Critical polymers: paths meet on fractal dust","At criticality, paths forced to share zero-dim dust","Random measures drive uncountable path intersections on fractals","Diamond fractal criticality: polymer paths overlap on dust"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction assumes Theorem 5.2 from the companion preprint, which asserts the existence and uniqueness of the edge-labeled random arrays with the prescribed variances and exact recursion; if that theorem or the variance-function lemma from [9] fails, the random measures $M_r$ need not exist.","fun_headline_variants_meta":{"raw":{"variants":["Critical polymers: paths meet on fractal dust","At criticality, paths forced to share zero-dim dust","Random measures drive uncountable path intersections on fractals","Diamond fractal criticality: polymer paths overlap on dust"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000512,"raw_usage":{"total_tokens":2559,"prompt_tokens":1085,"completion_tokens":1474,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":1411}},"tokens_in":701,"tokens_out":1474,"duration_ms":12749,"temperature":1.0,"reasoning_tokens":1411,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:25:30.818272+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the finite-$n$ pair correlation $\\mathbb{E}[M_{r,n}(p)M_{r,n}(q)]$ for two cylinder paths in the discrete model of Theorem 2.22; Lemma 2.7 predicts it equals $|\\Gamma_n|^{-2}(1+R(r-n))^{\\xi_n(p,q)}$. A measured deviation from this formula for large $n$, or a direct construction of two different array laws satisfying Theorem 5.2 for the same $r$, would disprove the uniqueness or existence claim.","supporting_citations":[{"cited_title":"Clark: High-temperature scaling limit for directed polymer on a hierarchical lattice with bond disorder, J","cited_arxiv_id":null,"evidence_quote":"Supplies Lemma 2.5, the unique variance function $R(r)$ with its recursion and asymptotics, which underpins all variance and correlation formulas."},{"cited_title":"The conditional Gaussian multiplicative chaos structure underlying a critical continuum random polymer model on a diamond fractal","cited_arxiv_id":"1908.08192","evidence_quote":"Provides the proof that the total mass $M_r(\\Gamma)$ converges to zero in probability as $r\\to\\infty$ and the conditional GMC representation used in Proposition 2.16."},{"cited_title":"Hausdorff: Dimension und ueres Ma , Math","cited_arxiv_id":null,"evidence_quote":"Defines generalized Hausdorff measures used to quantify the zero-dimensional intersection sets and to state the log-Hausdorff exponent results."}],"review_version":1}