{"id":"a7821daa-29aa-461b-83d0-9198682b34b3","arxiv_id":"2504.13743","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The frontier of a simple planar random walk, parameterized by the 4/3-dimensional Minkowski content of the Brownian frontier, converges weakly to the Brownian frontier in the natural-parametrization metric.","lead":"This paper proves that the outer boundary of a planar random walk, when drawn at the correct fractal speed, converges to the outer boundary of planar Brownian motion, including how much time is spent in each region. For probability theory, it settles a natural open question about the strongest form of convergence of random walk frontiers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The L2 estimate (5.13) requires Theorem 4.1(ii), whose proof is omitted; with Theorems 3.2–3.3 and (2.12) also deferred, Theorem 1.2 and hence Theorem 1.1 are unsupported as submitted.","rationale":"I read the paper in good faith: it aims to prove natural-parametrization convergence of the random walk frontier, with occupation-measure convergence as the intermediate theorem. The strategy is standard and the appendices contain real proof content, but the submitted text does not provide enough of the proof to verify the central claim. My concern is the same family as the reader's weakest assumption but more specific: Theorem 4.1(ii) is a sharp two-point estimate stated without any proof, whereas the paper explicitly proves only Theorem 4.1(i). This estimate is directly used, together with Proposition 4.9 and Lemma 5.5, to obtain the L2 bound (5.13). Without a proof of Theorem 4.1(ii), the second moment of the discrete occupation measure is uncontrolled, so the convergence of ν_n to ν is not established. I also flag that the one-point result depends on (2.12), whose constant q is left unspecified and whose proof is omitted, and that Theorems 3.2 and 3.3 are deferred to companion papers. If any of these sharp estimates carries only a polynomial error instead of the asserted e^{-un} error, the occupation-measure convergence and the natural-parametrization identification break. Since the reader's verdict is already conditional and my read does not move it, I recommend no change: the paper should remain conditional pending full proofs or precise verifiable references for these estimates.","tokens_in":30629,"tokens_out":13968,"duration_ms":120230,"concrete_test":"Write out the full proof of Theorem 4.1(ii) following the template of part (i), or locate it verbatim in [3] or [5]. As a minimum check, derive the two-point analogue of Proposition 4.4 and verify that the constant in (4.9) is uniform in z,w for |z−w| ≥ e^{-n/6} and that the leftover cross-term is O(e^{-un}) times the leading term. If the proof yields only P(A_n(z) ∩ A_n(w)) ≍_V |z−w|^{-α} e^{-2αn} up to constants, then (5.13) cannot be concluded and Theorem 1.2's L2 convergence is unsupported; the natural-parametrization identification would then fail.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is Theorem 1.1, which is derived from Theorem 1.2 through the L2 estimate (5.13). That estimate is the load-bearing wall. Its discrete half requires the sharp two-point bound Theorem 4.1(ii): c1^2 e^{2αn} P(A_n(z) ∩ A_n(w)) ≃_V G^fr_D(z,w). The text proves only part (i) after Theorem 4.1; the proof of part (ii) never appears, and the preamble 'To derive Theorem 4.1...' does not supply it. The omitted proof would need a two-point version of Proposition 4.4 with a uniform constant and an exponentially small cross-term; any hidden dependence on |z−w| or any polynomial error would void (5.13). Moreover, Theorem 4.1(i) itself uses (2.12), a sharp non-disconnection estimate whose constant q is unspecified and whose proof is 'omitted for brevity'; the constant c1 is defined as 1/(c2 q), so the main theorem's normalization inherits an unproven constant. The continuum inputs, Theorems 3.2 and 3.3, are likewise deferred to cut-point analogues in [5] and [3]. Each of these is used with e^{-un} error; the paper gives no way to see that the frontier analogues preserve the exponential rate rather than a polynomial one. Since (5.13) feeds Theorem 6.5 (tightness under ρ) and Theorem 6.7 (uniqueness of subsequential limits), any weakening of these estimates leaves Theorem 1.1 unproven.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims two main results. Theorem 1.1 states that the frontier of a simple random walk on the scaled lattice Z_n = e^{-n}Z^2, traversed with each edge taking time c_1 e^{-4n/3}, converges weakly under the natural parametrization metric rho to the frontier of planar Brownian motion, with the Brownian frontier parameterized by its 4/3-Minkowski content measure. Theorem 1.2 states that the associated renormalized occupation measure nu_n = c_1 e^{-4n/3} sum_{x in Z_n cap fr(lambda_n)} delta_x converges weakly to the Minkowski content measure nu of the Brownian frontier. The proof strategy is to establish one- and two-point frontier Green's function estimates, prove an L2 approximation between nu_n and nu, use this to derive convergence of occupation measures, and then combine this with convergence modulo reparametrization, tightness under rho, and uniqueness of subsequential limits. The overall architecture follows the cut-point program of the authors' earlier work and uses Skorokhod and KMT couplings, non-disconnection exponents, quasi-invariant measures, and Aizenman-Burchard tightness.","tokens_in":31265,"tokens_out":10370,"duration_ms":92444,"significance":"If the missing estimates are supplied, the paper would be a significant advance: it upgrades the known Hausdorff-metric convergence of random walk frontiers to convergence in the stronger natural-parametrization metric and identifies the normalization constant c_1 in terms of the two-arm disconnection exponent. The paper's strategy is coherent, and the appendices do contain detailed proofs of several substantial propositions, including Propositions 4.2, 4.3, and 4.4 and Lemma D.1; the use of thickenings and quasi-invariant measures is appropriate. However, the central load-bearing statements are not proved in this manuscript: the sharp frontier Green's function estimates, the sharp SRW non-disconnection estimate, the inward coupling theorem, the discrete-continuum event comparison lemma, the two-point moment identification, and the final L2 estimate are either stated without proof or only sketched. As submitted, the main theorems are therefore conditional on a large number of unproved inputs.","major_comments":[{"comment":"The proof of Theorem 4.1 proves only part (i). Part (ii), the two-point estimate c_1^2 e^{2 alpha n} P(A_n(z) cap A_n(w)) simeq_V G^fr_D(z,w), is never established. There is no two-point analogue of Proposition 4.4 or Proposition 4.5, no discussion of the uniformity of the implied constant in V, and no proof that the cross-terms are exponentially small. This estimate is directly used in the derivation of (5.13) in Section 5.2, which is the engine of Theorem 1.2 and, through it, Theorem 6.7. Without a proof of part (ii), the two-point moment identification underlying convergence of the occupation measure is unsupported.","section":"Section 4.1, Theorem 4.1(ii)"},{"comment":"The sharp SRW non-disconnection estimate P_{0,0}(D_m) simeq q e^{-alpha m} is asserted without proof, with the text saying the proof is 'omitted for brevity' and with the constant q left unspecified. This is load-bearing because the proof of Theorem 4.1(i) defines c_1 = 1/(c_2 q), so the normalization of nu_n in Theorem 1.2 and the parametrization speed in Theorem 1.1 inherit this unproved constant. The exponential error in (2.12) is also used throughout the moment computations. A complete proof or a precise published reference for this sharp estimate is necessary.","section":"Section 2.2, Eq. (2.12)"},{"comment":"The one- and two-point frontier Green's function estimates, the existence of the non-atomic measure nu with moments given by (3.8), and the L2 estimate (3.9) for the continuum frontier-disk measure are all stated as theorems but their proofs are omitted. The text says the proofs follow 'similarly as in their cut-point counterparts with minor technical modifications', but these are not minor or local results: Theorem 3.3 is used in the two-point estimate and Theorem 3.5 is used directly in the continuum half of (5.13). If any of these statements has only polynomial error instead of the asserted e^{-us} error, the L2 convergence and hence the natural-parametrization identification break. These theorems need either full proofs or a precise statement of the dependence on the cited works [3] and [5].","section":"Section 3, Theorems 3.2-3.5"},{"comment":"Theorem 4.6 (inward coupling for non-disconnecting walks) and Lemma D.2, which is said to imply Proposition 4.5, are both stated without proof. Theorem 4.6 is used essentially in Lemma 4.10 and hence in Propositions 4.8 and 4.9, while Lemma D.2 underlies the comparison in Proposition 4.5 that enters the proof of Theorem 4.1. The statements themselves are nontrivial, and the required exponential error rate e^{-u(n-m)} or simeq cannot simply be assumed from analogous cut-point results. Omitted proofs of these two statements leave the discrete-to-continuum comparison unverified at the precision needed by the paper.","section":"Sections 4.2 and Appendix D"},{"comment":"The chain of equivalences in (4.11) uses Proposition 4.2 to replace the sum over Y_{n/6}(z_n) by the sum over NICE_{n/6}(z_n). Proposition 4.2 is stated for n >= m >= 10l. In this application l = n/6, so the condition requires m >= 10l = 5n/3, which is incompatible with m <= n. Thus the stated Proposition 4.2 cannot be applied at the scale needed in (4.11). Either Proposition 4.2 needs a different formulation, or the proof of Theorem 4.1(i) needs a different scale choice; as written, the replacement of Y_{n/6} by NICE_{n/6} is unjustified.","section":"Section 4.1, proof of Theorem 4.1(i)"},{"comment":"The proof of tightness under rho asserts that nu_n(V) -> nu(V) in probability for any rational disk V and then invokes (5.13). However, (5.13) is proved only for nice boxes V in V. The paper does not prove the extension from nice boxes to arbitrary rational disks, although such an extension may be obtainable by approximating disks by nice boxes and using non-atomicity of nu. As written, this is a gap in the proof of Theorem 6.5 and therefore in the proof of Theorem 1.1.","section":"Section 6.2.1, Theorem 6.5"}],"minor_comments":[{"comment":"The metric rho as defined appears asymmetric in the two curves: the infimum is taken only over bijections alpha : [0,t_gamma] -> [0,t_{gamma'}], with no analogous reparametrization of gamma'. The natural parametrization metric usually involves reparametrizations of both curves; the formula should be corrected or clarified.","section":"Eq. (1.3)"},{"comment":"The constant q is said in (2.8) to depend on the initial configuration beta, while in (2.12) it is called simply 'a constant' for the origin-to-origin case. The paper later uses this q in defining c_1 without discussing whether the same constant applies after translation to z_n; the domain is bounded, so translation invariance is not automatic and the use of a single q should be justified.","section":"Section 2.2, (2.8) and (2.12)"},{"comment":"The notation D_{-s}(z) and D_{-2s/3}(z) for disks is nonstandard and is never explicitly defined; the reader must infer that these are disks of radius e^{-s} and e^{-2s/3} respectively.","section":"Section 3.1, Definition 3.1"},{"comment":"In the discussion after (C.1), 'total variance norm' should be 'total variation norm'.","section":"Appendix C"},{"comment":"The paper says 'we only provide a proof sketch' and omits the approximation of continuous test functions by step functions in the proof of (5.12). This is presented as following 'the same lines' as in [3]; for a self-contained proof, at least the key steps of that approximation should be included.","section":"Section 5.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's main theorems are conditional on a large number of deferred proofs: Theorems 3.2-3.5, Eq. (2.12), Theorem 4.1(ii), Theorem 4.6, Lemma D.2, and the derivation of (5.13). In addition, the proof of Theorem 4.1(i) appears to apply Proposition 4.2 outside its stated scale range. These are not minor omissions but the core analytic inputs of the paper. I recommend major revision and would want to see complete proofs or a clearly separated list of imported theorems with precise statements and references before considering the paper for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main theorem is the right one: strong convergence of the random walk frontier in natural parametrization. The paper closes a genuine gap between Hausdorff convergence of frontiers and the continuum SLE natural parametrization, and the intermediate occupation measure convergence is also new. Credit where due: the paper actually proves Theorem 4.1(i) and gives detailed path-decomposition arguments in Appendices A–C, and the overall strategy is sensible.\n\nBut the proof as submitted is not complete enough for me to verify. The load-bearing wall is the L2 estimate (5.13), which requires Theorem 4.1(ii) — the sharp two-point frontier Green's function estimate. Theorem 4.1(ii) is stated but never proved. The text says to follow the same lines as the proof of (i), but that does not supply the uniform cross-term estimate or the |z−w| dependence; any hidden polynomial error would break (5.13), and hence both Theorem 1.2 and Theorem 1.1. The stress-test note is correct on this point.\n\nThe same pattern repeats elsewhere. Theorems 3.2–3.5 (one- and two-point frontier Green's functions, existence of the measure, L2 approximation) are all stated with 'proofs are omitted' and left to analogues in [5] and [3]. Theorem 4.6, the inward coupling, is stated without proof. Lemma D.2, which underpins Proposition 4.5, is stated without proof. And the sharp non-disconnection estimate (2.12) — which fixes the constant q — is asserted with q unspecified and 'omitted for brevity'. Since c1 = 1/(c2 q), the entire normalization of the natural parametrization rests on an unproven constant. These are not cosmetic omissions; they are the estimates that make the argument work.\n\nI should say what is not a problem: there is no circularity. c1 is defined from the one-point asymptotics, not chosen to force the conclusion. The reliance on [3] is heavy, and [3] is an unpublished preprint by overlapping authors, which makes referee checking harder, but self-citation is not itself a flaw.\n\nMy honest guess is the theorem is true and the strategy works; the paper is written by people who know the L2 approximation method. But as submitted, the main claim is unsupported at key technical points. A referee could not certify it from the text alone.\n\nThis paper is for specialists, and it deserves a serious referee — the result is important enough that the community needs the proof on the record. My recommendation: send it to peer review, but with a clear demand for major revision: supply the missing proofs (at least Theorem 4.1(ii), Theorems 3.2–3.5 with verifiable details, and (2.12) with the constant specified), or precise pointers to where they can be checked, before acceptance.","headline":"The paper claims the long-awaited natural parametrization convergence of the random walk frontier, but several load-bearing estimates are stated without proof, so the central theorem cannot be verified from the text as submitted.","tokens_in":31528,"tokens_out":4548,"would_cite":false,"duration_ms":41351,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60J65","60G50","60F17","28A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"The random-walk frontier converges to the Brownian frontier in the natural parametrization metric, with time set by 4/3-Minkowski content.","keywords":["random walk frontier","Brownian frontier","natural parametrization","Minkowski content","occupation measure","disconnection exponent","frontier Green's function","scaling limit"],"falsifier":"Compute the ratio P(D_m)$e^{{2m/3}}$ for two independent simple random walks that keep the origin on the frontier; if for large m the ratio wanders by more than $e^{{-um}}$ (for example, decays like $m^{{-c}}$), then (2.12) fails and Theorem 1.2 loses its key input. Alternatively, check the frontier Green's function identity E[L_s(z)] ≃ G_fr_D(z) near |z| = 1/2: a polynomial rather than exponential error would invalidate Theorem 3.2 and the L2 estimate (3.9).","tokens_in":30443,"feed_emoji":"🌀","tokens_out":7303,"duration_ms":63107,"temperature":0.7,"pith_summary":"This paper claims that the outer boundary of planar simple random walk, drawn on the lattice $e^{{-n}}$$Z^{2}$ and run until it exits the unit disk, converges weakly as a parametrized curve to the outer boundary of planar Brownian motion. The convergence is in the natural parametrization metric, which compares both the shape of a curve and the way time is assigned along it. The time parameter on the Brownian side is the 4/3-Minkowski content of the frontier, so the result says the random walk's own traversal speed—each edge taking time c1 $e^{{-4n/3}}$—converges to the natural fractal clock of the Brownian frontier. The intermediate claim is that the renormalized occupation measure on random-walk frontier points converges to the 4/3-Minkowski content measure. A sympathetic reader would care because this upgrades the known Hausdorff-metric convergence of frontiers to a genuine curve-level scaling limit, matching the conjectured 4/3-dimensional structure of the Brownian boundary.","feed_headline":"Frontier of random walk converges to Brownian frontier in natural time","feed_subtitle":"The lattice walk's boundary matches Brownian's, speed and all: 4/3-dimensional time.","key_machinery":"The central objects are the frontier of a lattice path and the frontier-disk events: a small disk is a frontier disk if the path enters it, the two arms around it do not disconnect it from infinity, and the intermediate excursion stays close by. The one-point and two-point frontier Green's functions G_fr_D(z) and G_fr_D(z,w) are the limits of the normalized expected weights of these events, and they supply the first and second moments needed for the L2 estimate. The sharp non-disconnection estimate P_{0,0}(D_m) ≃ q $e^{{-αm}}$ with α=2/3, transferred from Brownian motion to simple random walk, is the quantitative input that fixes the constant c1 and the exponential error. The L2-approximation then compares the discrete occupation measure against the continuum frontier-disk measure, and the 4/3-Minkowski content measure ν serves as the time parameter of the limiting curve.","core_discovery":"Theorem 1.1 asserts that γ_n converges weakly to eγ under the natural parametrization metric ρ, where eγ traces the frontier of Brownian motion run until it first exits the unit disk and is parameterized so that Cont_{4/3}(eγ[0,t]) = t. Theorem 1.2 gives the equivalent measure statement: with ν_n = c1 $e^{{-4n/3}}$ Σ_{x∈Z_n∩fr(λ_n)} δ_x and ν(·) = Cont_{4/3}(·∩eγ), the laws of ν_n converge weakly to ν. The proof first establishes one- and two-point frontier Green's function estimates and sharp frontier-disk probabilities, uses an L2-approximation with a Skorokhod embedding to compare discrete and continuum frontier disks, and then turns the occupation-measure convergence into a natural-parametrization statement by proving tightness and uniqueness of subsequential limits. In the author's framing, the occupation-measure convergence is not an add-on; it is the mechanism that forces the time parametrization of the limiting curve to be the 4/3-Minkowski content measure.","pith_inferences":["If the deferred sharp estimates are completed as stated, the same L2 scheme should transfer to other lattice curves whose boundaries admit analogous one- and two-point Green's function estimates, such as killed or conditioned random walks in domains.","A direct numerical check of P(D_m)e^{2m/3} for moderate m would give an early indication of whether the exponential error in (2.12) is plausible before a full proof appears.","The method suggests that for any scale-invariant planar curve with the same two-arm disconnection exponent, the natural time should be its 4/3-Minkowski content; a testable extension is whether the frontiers of other lattice models obey the same occupation-measure limit."],"forward_implications":["The natural parametrization of the Brownian frontier is the weak limit of the random walk frontier's own traversal time, so the exponent 4/3 appears as a traversal-time exponent for lattice frontiers.","The renormalized occupation measure of frontier points converges to a non-atomic measure, so along the limiting curve no point is visited for a positive amount of time, ruling out 'stuck' subsequential limits.","The frontier Green's function estimates give an SLE-free route to the existence of the 4/3-Minkowski content of the Brownian frontier, as stated in Theorem 3.4.","Combined with the earlier Hausdorff-metric convergence, the result identifies the Brownian frontier as the unique subsequential limit of random walk frontiers under both the reparametrization metric and the natural parametrization metric."],"supporting_citations":[{"why":"Supplies the L2-approximation and occupation-measure convergence scheme for cut points that this paper adapts to frontiers.","marker":"[3]"},{"why":"Provides the one-point and two-point Green's function estimates and Minkowski-content framework for Brownian cut points, the model for Theorems 3.2 and 3.3.","marker":"[5]"},{"why":"Establishes Hausdorff-metric convergence of random walk frontiers to the Brownian frontier, used to identify subsequential limits under reparametrization.","marker":"[32]"},{"why":"Provides the tightness criterion for random curves that yields tightness of γ_n in the reparametrization metric.","marker":"[1]"},{"why":"Supplies the sharp Brownian non-intersection estimates, including (2.8), which underlie the sharp discrete estimate (2.12).","marker":"[23]"},{"why":"Provides the Skorokhod-embedding translation of sharp Brownian estimates to simple random walk, invoked for the omitted proof of (2.12).","marker":"[31]"},{"why":"Gives the up-to-constants discrete non-disconnection estimate (2.11) used throughout the moment estimates.","marker":"[16]"},{"why":"Establishes Minkowski content and natural parametrization for SLE curves, the continuum framework for the 4/3-content measure.","marker":"[17]"}],"fun_headline_variants":["Random walk frontier converges to Brownian under natural time","Frontier of random walk mirrors Brownian under natural parametrization","Natural parametrization ties random walk frontier to Brownian","Random walk boundary's natural time converges to Brownian's","Occupied measure drives random walk frontier to Brownian limit"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire proof rests on several sharp estimates—one-point and two-point frontier Green's functions and the exponential-error non-disconnection estimate for simple random walks—that are stated without full proofs in this paper; if any of them holds only with polynomial error, the occupation-measure convergence and the natural parametrization result collapse.","fun_headline_variants_meta":{"raw":{"variants":["Random walk frontier converges to Brownian under natural time","Frontier of random walk mirrors Brownian under natural parametrization","Natural parametrization ties random walk frontier to Brownian","Random walk boundary's natural time converges to Brownian's","Occupied measure drives random walk frontier to Brownian limit"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001125,"raw_usage":{"total_tokens":4605,"prompt_tokens":799,"completion_tokens":3806,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":415,"completion_tokens_details":{"reasoning_tokens":3724}},"tokens_in":415,"tokens_out":3806,"duration_ms":22887,"temperature":1.0,"reasoning_tokens":3724,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:02:26.595748+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the ratio P(D_m)$e^{{2m/3}}$ for two independent simple random walks that keep the origin on the frontier; if for large m the ratio wanders by more than $e^{{-um}}$ (for example, decays like $m^{{-c}}$), then (2.12) fails and Theorem 1.2 loses its key input. Alternatively, check the frontier Green's function identity E[L_s(z)] ≃ G_fr_D(z) near |z| = 1/2: a polynomial rather than exponential error would invalidate Theorem 3.2 and the L2 estimate (3.9).","supporting_citations":[{"cited_title":"Scaling limit of the occupation measure of random walk cut points","cited_arxiv_id":"2310.09592","evidence_quote":"Supplies the L2-approximation and occupation-measure convergence scheme for cut points that this paper adapts to frontiers."},{"cited_title":"Holden, G","cited_arxiv_id":null,"evidence_quote":"Provides the one-point and two-point Green's function estimates and Minkowski-content framework for Brownian cut points, the model for Theorems 3.2 and 3.3."},{"cited_title":"van de Brug, F","cited_arxiv_id":null,"evidence_quote":"Establishes Hausdorff-metric convergence of random walk frontiers to the Brownian frontier, used to identify subsequential limits under reparametrization."},{"cited_title":"Aizenman, A","cited_arxiv_id":null,"evidence_quote":"Provides the tightness criterion for random curves that yields tightness of γ_n in the reparametrization metric."},{"cited_title":"Shiraishi","cited_arxiv_id":null,"evidence_quote":"Provides the Skorokhod-embedding translation of sharp Brownian estimates to simple random walk, invoked for the omitted proof of (2.12)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the up-to-constants discrete non-disconnection estimate (2.11) used throughout the moment estimates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes Minkowski content and natural parametrization for SLE curves, the continuum framework for the 4/3-content measure."}],"review_version":1}