{"id":"4ce7c800-87ad-42d9-9648-7d6515ed782b","arxiv_id":"2607.15966","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Canonical scalar dyadic Mandelbrot cascades, and their pushforwards by fixed nondegenerate C² arcs and Jordan curves, are almost surely Rajchman under the minimal Kahane–Peyrière integrability conditions.","lead":"This paper proves that the Fourier transform of a canonical dyadic Mandelbrot cascade tends to zero at infinity, even in the heavy-tailed regime where no moment above order one exists. It also proves the same qualitative decay for the cascade pushed forward by any fixed smoothly curved arc or Jordan curve.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the proof's key inequalities are internally consistent and the curvature hypothesis is explicit, necessary, and used exactly as stated.","rationale":"The reader's weakest_assumption identifies the curvature nondegeneracy as load-bearing. I agree that it is structurally necessary and that the proof genuinely uses it to control derivative sublevel geometry and to exclude the affine resonance. However, I do not find this to be a correctness risk: it is an explicit, sharp hypothesis, and the manuscript's scope statement acknowledges its role. The part of the argument that is most likely to hide an error is the new probabilistic engine, especially the spine lower-deviation principle and the capped-martingale skeleton. I examined those sections and found the estimates internally consistent: Lemma 3.1 relies only on L^1-integrability of the size-biased increment and a dyadic truncation argument; Lemma 3.3 supplies a valid submartingale maximal bound; and Theorem 3.14 cleanly separates the uncapped large-jump and compensator terms, both of which are handled by L_n(theta) and by E[W 1_{W>t}]->0. The planar grid entropy is absorbed by Freedman's inequality after choosing the cap b/n small enough. The curve geometry in Section 4 appears sound, with the band count and phase-shift bounds uniform on each annulus. No circularity or overclaim was found. Thus the appropriate verdict remains ACCEPT/UNCHANGED, and the concrete test is a focused independent verification of the spine estimate, the least standard component.","tokens_in":36586,"tokens_out":40773,"duration_ms":419034,"concrete_test":"Have an independent expert re-derive Lemma 3.1 and Theorem 3.4 from the stated size-biased spine identities, or run a simulated cascade with a heavy-tailed W satisfying E[W log_2^+ W]<infinity and E[W^q]=infinity for q>1, checking that L_n(theta)->0 for n=2^10..2^20; a failure of either the summability of Psi_theta or the submartingale bound would invalidate Theorem 1.1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. The central claim rests on (i) the spine lower-deviation estimate Theorem 3.4, (ii) the terminal and stopped-skeleton controls Theorems 3.11 and 3.14, and (iii) the endpoint-safe curve decomposition in Proposition 4.8. I checked the load-bearing steps. Lemma 3.1's truncation/Kolmogorov argument is valid, and the two dyadic sums are bounded by 2E*|D|. Lemma 3.3's submartingale argument correctly converts the summability of Psi_theta into L_n(theta)->0. Lemma 3.5 is correct since min{1,2^{k-m}} <= 2^{k-m} and the finite initial sum decays. Theorem 3.14's three-way split (large jumps, predictable compensators, capped martingale) is coherent, and the Freedman exponent c n/b absorbs O(2^{2n}) grid entropy because b can be chosen arbitrarily small. The geometric lemmas in Section 4 provide uniform band control and phase-bin shifts with constants depending only on gamma and j. The nondegeneracy assumption inf |det(gamma',gamma'')|>0 enters exactly at Lemmas 4.1 and 4.2, and the affine-line obstruction (1.4) shows it is necessary rather than bookkeeping. The paper's stated limitations (fixed curve, no uniform event over curves, qualitative decay only) are honest scope statements. Companion papers enter only in Corollary 1.7, not in the main theorems.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that, under the minimal Kahane–Peyrière assumptions EW=1, E[Wlog_2^+W]<∞, and E[Wlog_2W]<1, the canonical scalar dyadic Mandelbrot cascade on [0,1] is Rajchman almost surely on non-extinction, and the same holds for the pushforward under any fixed nondegenerate C^2 embedded arc as well as for the circle cascade pushed forward by a fixed nondegenerate C^2 Jordan curve. The proof introduces a spine-based lower-deviation mass estimate (Theorem 3.4), a shifted-convolution consequence (Lemma 3.5), an adaptive terminal cutoff with terminal-descendant controls (Theorems 3.8, 3.11), an abstract stopped-skeleton theorem using Freedman/Bernstein capping (Theorem 3.14), and an endpoint-safe phase decomposition for nondegenerate arcs (Section 4, Proposition 4.8). The conclusions are qualitative: no positive power-law exponent is claimed. In the heavy-tailed regime E[W^q]=∞ for all q>1, the Rajchman theorems combine with exact Fourier-dimension formulas from companion papers to yield pure Rajchman measures of Fourier dimension zero (Corollary 1.7).","tokens_in":36871,"tokens_out":24968,"duration_ms":238517,"significance":"If correct, this settles the long-standing Rajchman question for canonical scalar Mandelbrot cascades at the minimal integrability threshold, completing the qualitative picture after the exact Fourier-dimension formulas. The main contribution is the construction of an annular decay mechanism that works without any L^{1+η} moment: the spine lower-deviation mass, the predictable capping of martingale increments, and the endpoint-safe phase decomposition are all new and potentially reusable. The curvature hypothesis is explicit and necessary: affine arcs are never Rajchman as planar measures by the normal-frequency obstruction (1.4). The paper honestly states that the exceptional null set may depend on the curve and that no uniformity over curves or polynomial rate is claimed. I checked the load-bearing steps—the lower-deviation mass theorem, the terminal-descendant theorem, the abstract stopped-skeleton theorem, and the endpoint-safe geometry—and found no derivation gap or circularity. The external dimension formulas enter only in Corollary 1.7 and are clearly separated from the main proofs.","major_comments":[],"minor_comments":[{"comment":"In the tail estimate, the bound ∑_{k=1}^{n+h} min{1,2^{k-m}} ≤ 2+h+C0 is valid only once n is large enough that m ≥ K0 for every admissible m (i.e. n ≥ K0+C0). Since the proof already says 'for all sufficiently large n', please state this additional condition explicitly to avoid a transient confusion.","section":"Lemma 3.5"},{"comment":"The sentence 'there exists c0=c0(θ)>0, independent of b, such that ... τ/|Γ_J| ≥ 2^{c0 n}' is slightly imprecise: for each fixed b>0 there is N_b and a positive c0 independent of b such that the bound holds for n≥N_b. The threshold necessarily depends on b, but the subsequent use only needs the rate to be fixed after b is chosen. Please rephrase.","section":"Theorem 3.14"},{"comment":"The symbol h is used both as the terminal offset integer (e.g. n+h) and as the angular function h_ξ(t)=|ξ·γ'(t)|/(|ξ||γ'(t)|). Although the context disambiguates, renaming one of the two (for instance using k for the offset in Section 4) would improve readability.","section":"Section 4"},{"comment":"The paper clearly states that the exact Fourier-dimension formulas from [5,6] are used only in Corollary 1.7. If the journal requires these companion preprints to be published, the references should be updated in the final version.","section":"Corollary 1.7"}],"recommendation":"minor_revision","confidential_remarks":"I found no mathematical reason to doubt the main theorems. The proof is long but internally consistent, and the load-bearing estimates are correctly derived. The minor comments concern only exposition and should be easy to address. I recommend acceptance after a brief revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this is a real result, not a repackaging. The paper proves that canonical scalar dyadic Mandelbrot cascades are Rajchman under only the Kahane–Peyrière assumptions, and extends to fixed nondegenerate C² arcs and Jordan curves. Prior work had exact Fourier dimension and polynomial decay under stronger moments; the zero-Fourier-dimension endpoint was open. The proof is long but the architecture is clear: spine lower-deviation mass, shifted convolution, adaptive terminal cutoff, stopped skeleton via Freedman, and an endpoint-safe phase decomposition. I checked the load-bearing inequalities: Lemma 3.1's truncation, Lemma 3.5's shifted convolution, Theorem 3.14's three-way split, and the geometric sublevel lemmas in Section 4. They are internally consistent. The curvature hypothesis inf|det(γ′,γ″)|>0 is genuinely necessary: affine arcs fail by (1.4), and it enters exactly in Lemma 4.1/4.2 to control stationary regions. The paper's stated limitations—fixed parametrization, no uniform event over all curves, no positive power law—are honest and correctly scoped. The self-citations enter only in Corollary 1.7, not in the main theorems.\n\nSoft spots are minor relative to the result. The proof is not machine-checked and has many moving parts; I would want a careful referee to verify the stopping-time argument and the Freedman constant absorption, especially in Theorem 3.14. The paper is dense but readable. The interval theorem is clean; the curved case adds real geometric complexity, and the endpoint-safe treatment of the small-derivative and endpoint regions looks correct. The main risk is not a found flaw but the usual long-proof risk of a subtle indexing or measurability issue. Nothing in the current text suggests one.\n\nWho is this for? People working on fractal harmonic analysis, Fourier decay of random measures, and Mandelbrot cascades. It deserves a serious referee. I would send it out and, if the proof survives refereeing, accept it. I would cite it in my own work on Fourier dimension of random measures.","headline":"This paper settles the Rajchman endpoint for canonical scalar cascades under minimal integrability, and the proof machinery looks sound.","tokens_in":37394,"tokens_out":1622,"would_cite":true,"duration_ms":17105,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G57","42B10","28A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"Canonical Mandelbrot cascades on intervals and on nondegenerate curves have Fourier transforms tending to zero at infinity, almost surely on non-extinction, under minimal Kahane–Peyrière assumptions.","keywords":["Mandelbrot cascades","Rajchman measures","Fourier decay","Fourier dimension","heavy tails","size-biased spine","nonvanishing curvature","Kahane–Peyrière conditions"],"falsifier":"Simulate a dyadic cascade with $W$ satisfying $E W = 1$, $E[W \\log_2 W] < 1$, and $E[W^q] = \\infty$ for all $q > 1$ (for example with a tail decaying like $c/(t \\log^2 t)$), generate a non-extinct realization, and compute $|\\hat{\\mu}(2^N)|$ for large $N$; if a persistent nonzero floor appears, Theorem 1.1 is false. The same check applied to $\\gamma(t) = (t, t^2 + t^3)$ along a fixed normal direction would test Theorem 1.4.","tokens_in":36431,"feed_emoji":"📉","tokens_out":7858,"duration_ms":82030,"temperature":0.7,"texified_at":"2026-08-05T21:26:12.022414+00:00","pith_summary":"Mandelbrot cascades are random multiplicative measures built on a binary tree; the paper asks whether their Fourier transform dies out at infinity, and answers yes at the weakest possible integrability threshold. It proves that the canonical scalar dyadic cascade on $[0,1]$ is Rajchman almost surely whenever it survives, assuming only the standard minimal conditions on the weight: mean one, finite $W \\log^+ W$, and logarithmic expectation below one, with no moment of order greater than one required. It then shows the same for the pushforward of the cascade by any fixed nondegenerate curved arc or Jordan curve in the plane, where the curvature prevents a permanent normal-frequency resonance. In the heavy-tailed regime where every higher moment is infinite, the exact Fourier-dimension formulas give Fourier dimension zero, so these theorems produce 'pure Rajchman' measures—decay at infinity with no positive power-law decay. A sympathetic reader would care because this settles the qualitative endpoint of a question Mandelbrot raised about turbulence spectra, using a mechanism that does not rely on any smoothness or higher moments.","texify_model":"deepseek-v4-flash","texify_usage":{"total_tokens":6203,"prompt_tokens":890,"completion_tokens":5313,"prompt_tokens_details":{"cached_tokens":0},"prompt_cache_hit_tokens":0,"prompt_cache_miss_tokens":890,"completion_tokens_details":{"reasoning_tokens":4438}},"feed_headline":"Mandelbrot cascade spectra vanish at infinity, heavy tails included","feed_subtitle":"Even with infinite moments, dyadic cascades and their curved images are Rajchman almost surely on non-extinction.","key_machinery":"The argument runs on three coupled components. First, a spine-based lower-deviation principle shows that the generation-$n$ mass carried by abnormally large cylinders, $L_n(\\theta)$, tends to zero almost surely for every $\\theta$ below the typical growth exponent $\\chi$, and that this survives the shifted-convolution sums over generations that Fourier analysis requires. Second, an adaptive terminal cutoff discards large terminal cylinders at a threshold of order $1/n$, while predictable capping turns the remaining martingale increments into a stopped skeleton whose concentration bounds absorb the entropy of a planar frequency grid. Third, an endpoint-safe phase decomposition for a nondegenerate arc uses the strict","core_discovery":"The paper's central claim is that the Rajchman property holds at the qualitative endpoint of the Mandelbrot cascade spectrum: for the canonical scalar dyadic cascade $\\mu$ on $[0,1]$ satisfying only $W \\ge 0$, $E W = 1$, $E[W \\log^+_2 W] < \\infty$, and $E[W \\log_2 W] < 1$, one has $\\hat{\\mu}(\\xi) \\to 0$ as $|\\xi| \\to \\infty$ almost surely on non-extinction (Theorem 1.1). For each fixed nondegenerate $C^2$ embedded arc $\\gamma$, meaning an embedding with speed and curvature bounded away from zero, the same holds for the pushforward $\\gamma_\\#\\mu$ (Theorem 1.4), and for each fixed nondegenerate $C^2$ Jordan curve $\\Gamma$, the circle cascade pushed forward by $\\Gamma$ is Rajchman (Theorem 1.6). These are not consequences of any positive power-law bound: in the regime $E[W^q]$","pith_inferences":["A testable extension is to push the same spine-gauge and capping routine through vector-valued and b-adic cascades; the geometric leg of the proof is already separated from the probabilistic leg and would not need to change.","The necessity of the curvature condition suggests an open boundary problem: arcs whose curvature vanishes at isolated points, so that the tangent angle is only weakly monotone, fall outside the theorem but may still be Rajchman or not depending on the vanishing order.","Because the proof is entirely qualitative, it predicts no specific decay rate; if a rate exists, it must degrade as the tail of W becomes heavier, since the Fourier dimension drops to zero exactly in that regime."],"forward_implications":["The interval cascade's Fourier transform tends to zero along every unbounded frequency set, almost surely on the non-extinction event, under the minimal Kahane–Peyrière assumptions.","Every fixed nondegenerate C² arc pushforward and every fixed nondegenerate C² Jordan-curve pushforward is Rajchman almost surely on the corresponding non-extinction event.","In the heavy-tail regime E[W^q] = ∞ for all q > 1, the interval cascade and its curved pushforwards are pure Rajchman: Fourier dimension zero yet Fourier transform vanishing at infinity.","The decay is simultaneous over all large frequency annuli on a single almost-sure event, not merely along a lacunary sequence of frequencies.","The nonvanishing-curvature hypothesis is not removable: affine arcs violate it, and by equation (1.4) their pushforwards are never Rajchman."],"fun_headline_variants":["Rajchman decay proven for Mandelbrot cascades on curves","Curved cascades: Fourier coefficients vanish at infinity","Infinite moments allowed: curves keep cascade spectra vanishing","Mandelbrot cascades on arcs: Rajchman at minimal threshold","Pushforward cascades are Rajchman on any nondegenerate curve"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing assumption is geometric: the given arc or Jordan curve must have speed and curvature bounded away from zero ($\\inf |\\det(\\gamma', \\gamma'')| > 0$); if that fails, as for a straight segment, the pushforward is never Rajchman, so the proof's mechanism, not merely its constants, depends on it.","fun_headline_variants_meta":{"raw":{"variants":["Rajchman decay proven for Mandelbrot cascades on curves","Curved cascades: Fourier coefficients vanish at infinity","Infinite moments allowed: curves keep cascade spectra vanishing","Mandelbrot cascades on arcs: Rajchman at minimal threshold","Pushforward cascades are Rajchman on any nondegenerate curve"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000231,"raw_usage":{"total_tokens":1377,"prompt_tokens":856,"completion_tokens":521,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":432}},"tokens_in":600,"tokens_out":521,"duration_ms":4789,"temperature":1.0,"reasoning_tokens":432,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T21:45:51.013054+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a dyadic cascade with $W$ satisfying $E W = 1$, $E[W \\log_2 W] < 1$, and $E[W^q] = \\infty$ for all $q > 1$ (for example with a tail decaying like $c/(t \\log^2 t)$), generate a non-extinct realization, and compute $|\\hat{\\mu}(2^N)|$ for large $N$; if a persistent nonzero floor appears, Theorem 1.1 is false. The same check applied to $\\gamma(t) = (t, t^2 + t^3)$ along a fixed normal direction would test Theorem 1.4.","supporting_citations":[],"review_version":1}