{"id":"7942fbcd-988c-418b-83c5-3b310ff17515","arxiv_id":"2607.18160","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Every multi-parameter polynomial ergodic average on a measure space with commuting transformations converges almost everywhere; the proof gives sharp ℓ^p oscillation and maximal inequalities for discrete Radon averages.","lead":"The paper proves that multi-parameter polynomial ergodic averages converge pointwise for all commuting measure-preserving transformations, resolving the multi-parameter Bellow–Furstenberg problem. It introduces a multi-parameter circle method and a new Ionescu–Wainger multiplier theorem that yield quantitative maximal and oscillation estimates for discrete Radon averages.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Theorem 1.23 is written only for k=2; the asserted adaptation to k≥3, especially the small-scale variable-by-variable circle method and Lemma 7.61, is not demonstrated. The main theorem therefore rests on an unproved generalization.","rationale":"The paper is a serious and technically rich contribution: the multi-parameter Ionescu–Wainger theorem (Theorem 4.48), the large/small-scale decomposition in §5–§7, and the major-arcs rigidity counting lemma are substantial and internally plausible. I found no circularity or data-fitting; the citations to prior work, especially [9], [34], [35], and [50], are to independent results. The central vulnerability is exactly the gap the reader identified: Theorem 1.23 is stated for all k but proved only for k=2. The manuscript flags this itself, which is honest, but a proof cannot simply assert that a parameter-induction is 'readily seen' when the displayed argument in Lemma 7.61 is explicitly one-variable (Vandermonde in m1) and the surrounding circle method is built on a two-scale hierarchy. The k≥3 generalization is load-bearing because Theorem 1.7 is the headline result and it quantifies over all k; without the adaptation, the paper's written proof supports only k=2. The recommended verdict is therefore CONDITIONAL: the main claim should be accepted only after the k≥3 adaptation is supplied, or the theorem should be restated with k=2 and a separate assertion for k≥3.","tokens_in":82088,"tokens_out":7477,"duration_ms":84172,"concrete_test":"Independently write out the proof of Theorem 1.23 for k=3 for the monomial set Γ={(2,3,4),(1,2,5)} (or any rank-2 set), following §§5–7. In particular, prove the small-scale inequality (5.20) in the regime M2≤M3^ϱ, M1≤M2^ϱ. This requires: (i) the two-variable analogue of Lemma 7.61: Σ_{m1∈[M1], m2∈[M2]} 1_{A^{m1,m2}_{M1,M2}}(ξ) ≲ (M1M2)^{-4α} uniformly in the relevant denominators; (ii) a version of the major-arc approximation (7.80) with Ψ^{(Γ2)} depending on (m1,m2); (iii) the summation-by-parts step in §7.2 after replacing the one-variable polynomial ξ^{Γ2}(m1) by a bivariate one. If (i) yields only M1^{-4α} with no saving in M2, or requires M2 ≤ M1^{c} instead of the stated hierarchy, then Theorem 1.23 for k=3 is not established as written; if (i)–(iii) go through, the adaptation is routine and the concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claim (Theorem 1.7, all d,k∈Z+) is deduced from Theorem 1.23. But the manuscript explicitly proves Theorem 1.23 only for k=2: §1.3 says 'we prove Theorem 1.23 in the two-parameter setting k=2... a careful reader will readily see...', and §5 repeats 'For ease of exposition, we only prove Theorem 1.23 in the two-parameter setting k=2.' This is not a cosmetic omission: the k=2 proof is organized around a two-scale hierarchy. In §7, inequality (5.20) treats the small-scale regime M1≤M2^ϱ by applying the circle method first in the larger variable m2. The coefficient of m2^{γ2} is the one-variable polynomial ξ^{Γ2}_{γ2}(m1)=Σ ξ_{(γ1,γ2)}m1^{γ1}, and Lemma 7.61 counts m1 using a (d1+1)×(d1+1) Vandermonde determinant. For k=3, after the first circle-method step the remaining coefficient is a bivariate polynomial in (m1,m2); to iterate one would need a two-variable counting lemma with a saving that is uniform in both smaller scales, and the 'rank/parameters-gluing' reduction of Lemma 4.45 only lowers k to r≤min(d,k), not to 1. No such statement is formulated, and the asserted 'readily see' adaptation is not a proof. Consequently the written argument establishes Theorem 1.23 (and hence Theorem 1.2/1.7) only for k=2 unless the adaptation is supplied. The coefficient-1 lifting (Remark 7, §1.3, via [50, §1.4]) is a second assumed reduction, but the k≥3 gap is the primary obstacle.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a multi-parameter circle method for discrete Radon averaging operators on Z^d, in the framework of discrete analogues in harmonic analysis. Its main quantitative result is Theorem 1.23, a long multi-parameter oscillation inequality for truncated polynomial Radon averages; from this the authors deduce Theorem 1.2 (maximal and oscillation inequalities for the untruncated averages) and then, via Calderón transference, Theorem 1.7 asserting the affirmative solution of the multi-parameter Bellow–Furstenberg problem for all d,k. The key technical novelties are a multi-parameter Ionescu–Wainger theorem (Theorem 4.48), an iterative variable-by-variable circle method for the small-scale regime, and a new 'major-arcs rigidity' phenomenon captured in Lemma 7.61. The written proof, however, is carried out only for k=2: Section 1.3 states that Theorem 1.23 is proved in the two-parameter setting and that the extension to k>=2 is left to the reader, and Section 5 repeats this declaration.","tokens_in":82558,"tokens_out":4361,"duration_ms":51287,"significance":"If the results are correct in the stated generality, they resolve a longstanding open problem in ergodic theory and provide the first quantitative maximal and oscillation estimates for genuinely multi-parameter discrete Radon averages with polynomial mappings. The paper also introduces substantial technical machinery—the multi-parameter Ionescu–Wainger theory and the major-arcs rigidity phenomenon—that is likely to have independent value. The sharp parameter ranges, the coefficient-independence of the constants, and the clean reduction from ergodic theory to integer-shift estimates are all valuable features. The significance would be very high, provided the proof is supplied for arbitrary k>=3.","major_comments":[{"comment":"The reduction to canonical monomials with coefficient 1 is asserted rather than proved: 'This is a simple consequence of the lifting procedure, which can, for instance, be adapted from [50, Section 1.4].' This reduction is used throughout Sections 5–7, where the multiplier is written for the canonical mapping (t)^Gamma. If this lifting step is not exactly as in [50]—which is a one-parameter jump-inequality paper—the main theorem is not proved even for k=2. The manuscript should include a proof or a precise statement of the adapted lifting procedure.","section":"§1.3, Remark 7 after Theorem 1.23"},{"comment":"Theorem 4.15 is stated as a theorem of this paper, but its proof is deferred with one sentence: 'The proof proceeds in essentially the same way as the proof of [9, Theorem 6.14].' This result is used in crucial places, for example in §7.1.5 and §7.1.11. Since Theorem 4.15 is not literally the cited theorem but a one-parameter seminorm variant of Theorem 4.4, the paper should either provide the reduction to [9, Theorem 6.14] in detail or explicitly state that the theorem is imported verbatim from [9].","section":"§4.2, Theorem 4.15"}],"minor_comments":[{"comment":"The text has many formatting errors: missing spaces in phrases such as 'Radonaveraging', 'multi-parameterBellow–Furstenberg', 'variablebyvariable', and similar. These should be corrected.","section":"Throughout"},{"comment":"Section 3 is explicitly restricted to k=2 ('we will focus on the two-parameter setting k=2, noting that our arguments can be adapted'). This is a consequence of the main gap, but the exponential-sum estimates for k>=3 are not formulated. If the extension to arbitrary k is retained, the relevant Weyl-type statements for k>=3 should be stated.","section":"§3"},{"comment":"The sharpness of the p-ranges in (1.3) and (1.4) is asserted but not proved or referred to a specific counterexample. Since the paper claims sharpness, a reference or argument should be provided.","section":"§1.2, Theorem 1.2"},{"comment":"The proof of Theorem 4.48 contains several omissions: Step 7 says 'The rest of the proof is almost identical... We omit the details', and Step 6 refers to [34, Inequality (3.11)] for a key Littlewood–Paley estimate. These may be acceptable, but they should be made explicit enough for the reader to verify.","section":"§4.4, Theorem 4.48"},{"comment":"Reference [76] is listed as 'Private communication. May 2024.' This is not a verifiable published source; if the result of Wierdl is used, a published or preprint reference should be supplied.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"I believe the paper is very likely correct in its essential strategy and the k=2 proof is substantial. The main obstacle is that Theorem 1.23, and hence the claimed solution to the multi-parameter Bellow–Furstenberg problem, is not proved for k>=3. This is a fixable but central gap: the authors need to add the full multi-parameter induction or explicitly narrow the statement. I would be willing to look at a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a serious paper and the k=2 case looks like a real advance, but the headline theorem for all d,k is not proved in the text. The paper itself says it only proves Theorem 1.23 in the two-parameter setting. For k≥3 you get an assertion, not an argument. That is the main soft spot and it is load-bearing, since Theorem 1.7 is deduced from Theorem 1.23.\n\nWhat is actually new: a multi-parameter circle method tailored to oscillation and maximal semi-norms; Theorems 4.19 and 4.48, the multi-parameter Ionescu–Wainger variants; and the small-scale analysis with the \"major arcs rigidity\" counting lemma, Lemma 7.61. The treatment of k=2, d≥2 is genuinely new; the d=1, k≥2 case was already in [9]. Lemma 7.61 is argued in detail, and the surrounding section gives a real mechanism rather than a slogan. I found no circularity: the self-citations refer to independent prior results, and the proof is a forward derivation.\n\nWhere it is soft: the k≥3 gap is not cosmetic. The small-scale regime is organized around applying the circle method variable-by-variable. For k=2, after the first step the remaining coefficient is a one-variable polynomial, and the Vandermonde counting lemma applies. For k=3, after the first step you get a bivariate polynomial; you would need a two-variable counting lemma with a saving uniform in the two smaller scales. Lemma 4.45 only reduces k to r≤min(d,k), not to 1. No such statement is formulated. \"Careful reader will readily see\" is not a proof. Also assumed without proof: the reduction to coefficient-one monomials in Remark 7, the estimates (5.27) and (5.29), and the proof of Lemma 7.18 is only sketched. Sharpness of the p-ranges is asserted, not shown.\n\nIf the k≥3 adaptation works, this solves a long-open problem. If not, the k=2 theorem still stands as a substantial new result. This deserves a serious referee, with the instruction to demand the k≥3 argument or a revised statement.","headline":"Genuine breakthrough for k=2, but the all-k theorem rests on an asserted adaptation that is not in the paper.","tokens_in":83138,"tokens_out":2494,"would_cite":true,"duration_ms":24465,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37A30","42B25","42B20","11L07","11P55"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes the multi-parameter Bellow–Furstenberg problem: multi-parameter polynomial ergodic averages converge pointwise for all d, k and all L^p with p>1.","keywords":["multi-parameter discrete Radon averages","pointwise ergodic theorem","Bellow–Furstenberg problem","maximal inequalities","oscillation inequalities","circle method","Ionescu–Wainger multiplier theorem","major-arcs rigidity"],"falsifier":"Check whether the iterative small-scale argument of Section 7 can be repeated for k=3: take M_1, M_2 both small relative to M_3 and see whether Lemma 7.61 can be iterated to give a power saving in both smaller-scale variables. If the Vandermonde counting does not extend to two smaller-scale variables, then inequality (5.20) is not proved for k=3 and the main theorem as stated is unsupported.","tokens_in":81976,"feed_emoji":"🧮","tokens_out":5286,"duration_ms":99074,"temperature":0.7,"pith_summary":"The paper proves quantitative maximal and oscillation inequalities for discrete multi-parameter polynomial Radon averages, in which one averages a function over a box of polynomial images in Z^d. The bounds hold for every p>1 (and p=∞ for the maximal estimate), with constants that depend on the degree but not on the coefficients of the polynomials. These inequalities are proved using a new multi-parameter variant of the circle method, built on the Ionescu–Wainger multiplier theorem and an iterative, variable-by-variable argument in the small-scale regime. The main consequence is a positive resolution of the multi-parameter Bellow–Furstenberg problem in pointwise ergodic theory for all dimensions d and k. The written proof is carried out explicitly for k=2, with the general case left to an asserted adaptation.","feed_headline":"Polynomial ergodic averages converge pointwise for all d,k","feed_subtitle":"A multi-parameter circle method proves sharp maximal and oscillation bounds, resolving the Bellow–Furstenberg problem for every p>1.","key_machinery":"The central object is the truncated multi-parameter discrete Radon average and its Fourier multiplier, together with the oscillation seminorm (2.17) used to convert quantitative bounds into pointwise convergence. The key mechanism is an iterative application of the Hardy–Littlewood circle method, supported by two multi-parameter Ionescu–Wainger multiplier theorems (Theorems 4.19 and 4.48) that transfer continuous estimates to discrete settings with multi-frequency features. A rank parameter r, the rank of the d-by-k matrix of monomial exponents, tracks the number of independent 'directions'; when r<d, the 'parameters-gluing' obstruction is overcome by a counting lemma (Lemma 7.61) that estab","core_discovery":"The paper establishes that the multi-parameter oscillation inequality for truncated discrete polynomial Radon averages can be proved by iterating the classical circle method variable by variable. In the small-scale regime, this reveals a 'major-arcs rigidity' phenomenon: either the coefficient polynomial of the largest-scale variable lies on major arcs very rarely, yielding a power saving in the smaller-scale variables, or the full multi-parameter frequency lies on multi-parameter major arcs and the average is well approximated by a periodized continuous multiplier. Using this, the paper proves the sharp maximal and oscillation estimates (1.3) and (1.4), and derives the pointwise ergodic the","pith_inferences":["The authors leave implicit that the same approach should also yield the full (non-lacunary) oscillation inequality with constants independent of tau; they state this is possible but do not prove it.","If the k≥3 adaptation claimed in Section 1.3 holds, major-arcs rigidity is likely a general phenomenon: for multi-parameter polynomial phases, failure of joint major-arc structure forces power savings in the smaller-scale variables. A testable step is to generalize Lemma 7.61 to count m_1,...,m_{k-1} simultaneously.","The coefficient-independence of the constants rests in part on a monomial-lifting reduction that is assumed rather than proved; a direct verification of that lifting for general polynomial mappings would remove a hidden step in the argument."],"forward_implications":["For every polynomial mapping P with integer coefficients and any family of commuting invertible measure-preserving transformations, the averages A^P_{M;X,T}f converge pointwise and in L^p norm as min(M_1,...,M_k) tends to infinity, for all p>1.","The discrete Radon averages satisfy a maximal inequality for p in (1,∞] and an oscillation inequality for p in (1,∞), with sharp p-ranges and constants independent of the polynomial coefficients.","Calderón transference turns these integer-lattice estimates into quantitative maximal and oscillation ergodic theorems on arbitrary sigma-finite measure-preserving systems.","The multi-parameter Ionescu–Wainger theorems developed for oscillation seminorms are stated in sufficient generality to apply to other multipliers with comparable regularity, not only those arising from the Radon averages treated here."],"fun_headline_variants":["Multi-parameter Radon averages: sharp maximal bounds for all d,k","Pointwise ergodic theorem for all polynomials via new circle method","Major-arc rigidity proves pointwise convergence for all d,k","Sharp oscillation inequalities settle Bellow-Furstenberg for every p>1","Iterated circle method gives optimal multi-parameter Radon estimates"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof is written only for k=2, and the claim that all arguments adapt to arbitrary k≥3 is asserted in Section 1.3 but not demonstrated, so the theorem as stated for all k rests on that unproved adaptation, together with an assumed monomial-lifting reduction.","fun_headline_variants_meta":{"raw":{"variants":["Multi-parameter Radon averages: sharp maximal bounds for all d,k","Pointwise ergodic theorem for all polynomials via new circle method","Major-arc rigidity proves pointwise convergence for all d,k","Sharp oscillation inequalities settle Bellow-Furstenberg for every p>1","Iterated circle method gives optimal multi-parameter Radon estimates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001173,"raw_usage":{"total_tokens":4614,"prompt_tokens":595,"completion_tokens":4019,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":339,"completion_tokens_details":{"reasoning_tokens":3928}},"tokens_in":339,"tokens_out":4019,"duration_ms":26440,"temperature":1.0,"reasoning_tokens":3928,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:47:57.266814+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check whether the iterative small-scale argument of Section 7 can be repeated for k=3: take M_1, M_2 both small relative to M_3 and see whether Lemma 7.61 can be iterated to give a power saving in both smaller-scale variables. If the Vandermonde counting does not extend to two smaller-scale variables, then inequality (5.20) is not proved for k=3 and the main theorem as stated is unsupported.","supporting_citations":[],"review_version":1}