{"id":"114f9a62-b3a3-457d-8965-c0753890a95a","arxiv_id":"1908.04049","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Bochner-Riesz means of L^p functions converge almost everywhere on Heisenberg-type groups in a triangular range allowing arbitrarily small orders for some p bigger than 2.","lead":"This paper proves that Bochner-Riesz means of L^p functions converge almost everywhere on Heisenberg-type groups for a range of parameters. It extends a known theorem for Heisenberg groups to groups with higher-dimensional centers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 7.10's asserted θ-uniformity is not justified: the gamma factor grows polynomially in Im(a), so the stated estimate cannot hold uniformly in θ unless the e^{-θ²} factor from Proposition 7.3 is used explicitly.","rationale":"The reader correctly located the technical heart of the paper in the delicate estimates of Sections 7–8, but the most immediately load-bearing premise is not Theorem 8.4 itself. Before the Jacobi estimates are invoked, Lemma 7.10 must supply a uniform-in-θ kernel bound for complex orders. The proof's assertion that |C_{n,a}| depends only on Re(a) is contradicted by the standard asymptotics for ratios of gamma functions: for a = α−2iθ the ratio grows polynomially in θ. Since Proposition 7.3's hypothesis contains the factor e^{-θ²}, the polynomial growth is likely harmless, and the main theorem may still be true; however, the paper as written does not demonstrate this, so the conditional verdict is appropriate. The Jacobi-polynomial concern identified by the reader remains a secondary but genuine verification burden. The structural reductions (maximal-to-nonmaximal, interpolation, trace lemmas) are coherent and no circularity or data-fitting is apparent, so there is no reason to move to REJECT or to UNVERDICTED on the basis of this review.","tokens_in":51238,"tokens_out":18466,"duration_ms":175369,"concrete_test":"Re-derive (7.3) for ω = ψ and a = 2/3, tracking the factor |Γ(n/2−b/4)/Γ(b/4)| with b = 2/3 − 2iθ. Compute this ratio numerically for n = 1,2,3 at θ = 0,1,10,100 and verify that it grows like θ^{(n−2/3)/2}; then check whether multiplying by e^{-θ²} yields a θ-independent bound. If the damped bound holds, Lemma 7.10's assertion is simply missing a step and the proof can be repaired; if the damped bound fails, Proposition 7.3's hypothesis is not met and Theorem 7.1 is unproved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of the second-layer trace lemma uses Proposition 7.3, whose hypothesis (7.3) requires estimates for e^{-θ²}∂_{ω^{-a/2+iθ}}K uniformly in θ. In Lemma 7.10 this is reduced to bounding the kernel of ∂_{ψ^{-a/2}} for complex a. After formula (7.22), the proof claims |C_{n,a}| = π^{(Re(a)−n)/2}|Γ(n/2−a/4)/Γ(a/4)| ≲_{Re(a)} 1 and that 'the above quantity only depends on a through its real part.' This is false: for a = α−2iθ, the gamma ratio has magnitude ~ |θ|^{(n−α)/2} as |θ|→∞ (DLMF 5.11.12), with (n−α)/2 > 0 for α < min{2,n}. Hence Lemma 7.10's stated estimate (7.21), without the factor e^{-θ²}, cannot be uniform in θ. The complex-interpolation step that produces the non-radial estimate (7.4) therefore has a gap. The e^{-θ²} in (7.3) decays faster than any polynomial and may well repair the argument, but the repair is not written and the claim as stated is incorrect. This is upstream of the Jacobi estimates in Theorem 8.4 and affects the endpoint estimate (1.13), not merely the final interpolation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves an almost everywhere convergence result for Bochner–Riesz means of arbitrarily small order on Heisenberg-type groups, extending to H-type groups the range obtained by Gorges and Müller for Heisenberg groups. The proof has three main parts: a reduction of the maximal Bochner–Riesz operator to nonmaximal operators on certain weighted L2 spaces (Sections 4–6), a reduction of the relevant weighted estimates to dual Sobolev trace lemmas (Section 7), and the proof of those trace lemmas via refined estimates for Jacobi polynomials (Sections 7–8). The main theorem is stated as Theorem 1.1.","tokens_in":51489,"tokens_out":6119,"duration_ms":63764,"significance":"If the proof is completed, this is a substantial advance: it gives, on all H-type groups, a nontrivial range of p>2 for which almost everywhere convergence holds for every λ>0, with a sharp-looking trapezoidal range involving the parameter Q^*=2Q-D. The paper also introduces a clean maximal-to-nonmaximal reduction (inequality (1.19)/(4.10)) that is likely to be useful beyond this setting, and it develops explicit kernel formulas for fractional integration on the Fourier dual of H-type groups. The overall architecture is original and the debt to Gorges–Müller, Mauceri–Meda, and the Jacobi-polynomial literature is clearly acknowledged. However, the central technical lemma (Lemma 7.10) contains a gap: a uniformity claim in the imaginary part of the interpolation parameter is false as stated, and the proof as written does not establish the hypothesis (7.3) needed by Proposition 7.3. This gap is load-bearing for the main endpoint estimate (1.13).","major_comments":[{"comment":"Lemma 7.10 claims the uniform bound (7.21) for ∂_{ψ^{-a/2}}\\hat{K}_{γ,δ,j}, with no factor e^{-θ²}. For this to feed into Proposition 7.3, the bound is needed with a replaced by a-2iθ, since (7.3) involves ∂_{ω^{-a/2+iθ}}. In the proof, after formula (7.22), the constant is |C_{n,a}| = π^{(Re(a)-n)/2}|Γ(n/2-a/4)/Γ(a/4)|, and the text asserts that this is ≲_{Re(a)} 1 and that the quantity only depends on a through its real part. This is false: for a = α-2iθ (α = Re(a)), DLMF 5.11.12 gives |Γ(n/2-a/4)/Γ(a/4)| ≃ (1+|θ|)^{(n-α)/2}, which is unbounded in θ for 0<α<min{2,n}. Consequently the undamped quantity in (7.21) is not uniformly bounded in θ, and the stated lemma is incorrect. The gap is repairable in principle, because the e^{-θ²} factor in (7.3) decays faster than any polynomial and can compensate the growth; but the repair must be written. As it stands, the proof of (7.3) is missing, and Corollary 7.11, Theorem 7.1, and the endpoint estimate (1.13) all rely on it.","section":"§7.3, Lemma 7.10 and Proposition 7.3"},{"comment":"The proof of Theorem 8.4 invokes the asymptotic approximation of [21, eq. (3.49)] with the error bound from [6], displayed as (8.14). The text argues uniformity of the error by saying that ζ remains in a bounded interval, but the estimate must also be uniform in α,n in the regime 1≤α≤c(1+n), including the transition region where |ζ-α̃²| is comparable to α^{4/3}/u². This uniformity is used through the first estimate in (8.8) in the K1 and K3 estimates of Lemma 7.10, so the proof of the main trace lemma depends on it. The paper should either provide a detailed verification that the O(u^{-1}) error term in (8.14) is uniform in all parameters in the stated range, or cite a precise theorem covering exactly that range.","section":"§8, Theorem 8.4"}],"minor_comments":[{"comment":"The notation \"B(x,r)\" is used for both the open and closed ball in Section 2.1, which is confusing; please distinguish the two.","section":"Throughout"},{"comment":"The spelling \"Mihlin–Hörmander threshold\" should be \"Mikhlin–Hörmander\" to match standard transliteration.","section":"§1 and §3"},{"comment":"In the paragraph following (7.22), the paper states \"we may assume that a is real\" after asserting that the quantity depends only on Re a. This step requires the missing θ-dependence analysis described in the first major comment; as written it is not justified.","section":"§7.3, proof of Lemma 7.10"},{"comment":"In the proof of Theorem 8.4, the term \"E^{-1}_α M_α\" is used without defining the functions E_α and M_α in the manuscript; a brief definition or a precise pointer to [6] would improve readability.","section":"§8"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, this is a real extension: it gives the first trapezoidal-shaped a.e. convergence range for Bochner-Riesz means of a sub-Laplacian outside the Heisenberg setting, for all H-type groups. Second, read Lemma 7.10 with care; the claimed theta-uniform estimate has a gap, though I think it is repairable.\n\nThe main theorem is new and the paper does a lot well. The reduction from maximal to nonmaximal estimates is made explicit in (1.19), which is genuinely useful. The dual trace lemmas are the right technical core, and the paper is honest about its limitations: the exponent 2/3 is suboptimal, the case m=1 is deferred to Gorges-Muller, and several heuristic remarks are clearly labelled.\n\nNow the soft spots. The stress-test concern is legitimate: in Lemma 7.10, after formula (7.22), the constant C_{n,a} is not bounded uniformly in Im(a). Its magnitude grows like |theta|^{(n-Re(a))/2}, so the assertion that the quantity depends only on Re(a) is false. Consequently the estimate (7.21) cannot hold uniformly in theta as stated. The saving grace is that Proposition 7.3 asks for a bound on e^{-theta^2} times the kernel, and the Gaussian decay beats any polynomial in |theta|. So the theorem may survive, but only after inserting a factor |theta|^N in Lemma 7.10 and checking that it is absorbed. As written, this is a genuine gap, not a cosmetic one.\n\nA second point to watch is Theorem 8.4. The proof leans on external asymptotic expansions from Dunster and Boyd-Dunster, and the error term is asserted to be uniform in the stated parameter range. I did not find an obvious mistake, but a referee should verify the range 1 <= alpha <= c(1+n) actually covers all cases arising in (7.26). If it doesn't, the endpoint estimate (1.13) collapses.\n\nThe citation pattern looks honest: prior results are independent published theorems, and the self-citation to [13] is used for a known lemma. I found no circularity.\n\nBottom line: this paper deserves a serious referee. The main idea is new, the execution is thoughtful, and the identified gaps are likely fixable. I would engage with it in a reading group and cite it if I were working in this area.","headline":"Genuinely new trapezoidal-range a.e. convergence theorem for H-type groups, with a detailed proof that has one correctable uniformity gap in the second-layer trace lemma.","tokens_in":52070,"tokens_out":4101,"would_cite":true,"duration_ms":40215,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E30","43A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"Bochner–Riesz means converge almost everywhere on Heisenberg-type groups in a trapezoidal range.","keywords":["almost everywhere convergence","Bochner–Riesz means","Heisenberg-type groups","Jacobi polynomials","sub-Laplacian","maximal operator","dual Sobolev trace inequality","spectral multipliers"],"falsifier":"A concrete way to test the theorem is to evaluate the trace-lemma estimate (7.21) on a Heisenberg-type group with centre dimension $n>1$, in the middle-frequency range $\\frac34J_\\delta<j<J_\\delta$, using the explicit kernel formula (2.44) for the second-layer weight; if for a sequence $\\delta\\to0$ some test function violates the claimed bound with exponent $\\frac23$, the trace lemma is false and the theorem collapses. A more local check is to compute the two sides of (8.8) numerically for Jacobi parameters $\\alpha\\approx c(1+n)$ with $x$ near the transition point $x_{\\mathrm{tr}}$; any systematic violation would pinpoint the failing step.","tokens_in":50989,"feed_emoji":"📐","tokens_out":12376,"duration_ms":112261,"temperature":0.7,"pith_summary":"The paper establishes an almost-everywhere convergence theorem for Bochner–Riesz means of sub-Laplacians on Heisenberg-type groups. The theorem gives an explicit trapezoidal range of orders $\\lambda$ and exponents $p$ in which $T^\\lambda_r f\\to f$ almost everywhere as $r\\to 0^+$ for every $f\\in L^p(G)$, and the range contains some $p>2$ for every sufficiently small $\\lambda$. This matters because such a convergence range was previously known only for Heisenberg groups themselves; the paper extends it to the whole class of H-type groups, whose centre may have dimension larger than one. The proof is a chain of reductions: local maximal estimates for Bochner–Riesz means are controlled by non-maximal estimates, which are then reduced to dual Sobolev trace lemmas proved through refined Jacobi polynomial estimates.","feed_headline":"Bochner–Riesz means converge almost everywhere on Heisenberg-type groups","feed_subtitle":"The theorem yields p>2 for arbitrarily small smoothing orders, beyond the range known for general stratified groups.","key_machinery":"The proof is carried by two reduction steps. First, Lemma 3.1 and the estimates around (1.19) control the local maximal Bochner–Riesz operator pointwise by a product of two non-maximal operators, so the weighted $L^2$ estimate for the maximal operator is equivalent to uniform estimates for $m_\\delta(sL)$; the reduction is valid for sum-of-squares polynomial weights, and a separate interpolation step in Section 5 extends it to the fractional weights $(1+|\\cdot|)^a$ and $(1+\\rho)^b$. Second, those non-maximal estimates are recast as dual Sobolev trace lemmas, Theorems 7.1 and 7.2, which are weighted $L^2$ bounds for the spectral cut-offs $M^\\gamma_{\\delta,j}$ playing the role of a Sobolev trace inequality in frequency space. The trace lemmas are proved on the group-Fourier side: the kernel for fractional integration with respect to the second-layer weight $|u|$ is explicit in terms of Jacobi polynomials, and the decisive input is the refined Jacobi-polynomial estimate Theorem 8.4, whose transition point $x_{\\mathrm{tr}}=1-\\alpha^2/(2u^2)$ governs the two bounds in (8.8).","core_discovery":"The central claim is Theorem 1.1: if $G$ is an H-type group with homogeneous dimension $Q$, topological dimension $D$, and $Q^*=2Q-D$, then for every $\\lambda>0$ and every $2\\le p\\le\\infty$ satisfying $\\frac{Q^*-1}{Q^*}(\\frac12-\\frac{\\lambda}{D-1})<\\frac1p\\le\\frac12$, the Bochner–Riesz means $T^\\lambda_r$ converge almost everywhere to $f$ as $r\\to0^+$ for all $f\\in L^p(G)$. Because the left inequality admits positive solutions $\\frac1p$ for all small $\\lambda$, the admissible region is genuinely trapezoidal, so almost-everywhere convergence holds for some $p>2$ at arbitrarily small orders. The theorem covers all H-type groups, not only the Heisenberg groups for which an analogous result was already known.","pith_inferences":["If the missing trace estimate with exponent $a=1$ for the middle frequencies $\\frac34J_\\delta<j<J_\\delta$ could be established, the convergence range would match the known Heisenberg-group range; the natural route would be a weight mixing the $z$ and $u$ variables.","The same scheme may transfer to other 2-step stratified groups: the only group-specific ingredient is the Fourier-side kernel of the chosen weight, so finding a weight whose kernel obeys a suitable special-function estimate is a template for further results.","The appearance of $Q^*=2Q-D$ shows that the gain over the general stratified-group bound shrinks as the centre dimension grows; testing whether the trapezoid can be stated with $Q$ replaced by $D$ would distinguish a genuine geometric obstruction from an artifact of the trace-lemma method."],"forward_implications":["The theorem gives almost-everywhere convergence for every H-type group, extending the known Heisenberg-group result to centres of any dimension $n\\ge1$.","For every sufficiently small $\\lambda>0$ the admissible region contains exponents $p>2$, so convergence holds at arbitrarily small smoothing orders on these groups.","The explicit maximal-to-nonmaximal reduction (1.19) is a quantitative statement that does not appear in the Euclidean or Heisenberg-group precedents and can be applied to other maximal operators built from sub-Laplacians.","The dual trace lemmas supply the vertex estimate (1.13) at $\\frac1p=\\frac{Q^*-1}{2Q^*}$ and $\\lambda=0$, which is the endpoint needed for interpolation across the whole trapezoid.","Theorem 1.2 improves the general stratified-group maximal bound whenever the sub-Laplacian's spectral threshold $\\varsigma_+(L)$ is smaller than $Q/2$."],"supporting_citations":[{"why":"Supplies the Euclidean almost-everywhere-convergence template and the trace-lemma approach that this paper adapts.","marker":"[9]"},{"why":"Proves the Heisenberg-group case and contributes the reduction to non-maximal operators and fractional integration on the group-Fourier dual.","marker":"[30]"},{"why":"Provides the Jacobi-polynomial asymptotics used to prove the transition-point estimate Theorem 8.4.","marker":"[21]"},{"why":"Gives the uniform asymptotic theory behind the approximation in (8.14), with a turning point and a regular singularity.","marker":"[6]"},{"why":"Supplies the weighted Jacobi inequality used in the proof of Corollary 8.3.","marker":"[33]"},{"why":"Gives the refined Jacobi upper bound stated as Theorem 8.2(iii) and used in the large-degree regime.","marker":"[41]"},{"why":"Provides the Jacobi Bernstein-type inequality stated as Theorem 8.2(i).","marker":"[40]"},{"why":"Defines the spectral-multiplier threshold $\\varsigma_+(L)$ on which the basic estimates (1.15) and Theorem 1.2 rest.","marker":"[52]"},{"why":"Establishes the general stratified-group maximal Bochner–Riesz bound that Theorem 1.1 goes beyond.","marker":"[58]"}],"fun_headline_variants":["Arbitrarily small Bochner-Riesz order still gives a.e. convergence on H-type groups","A.e. convergence for Bochner-Riesz of any order on Heisenberg-type groups","H-type groups: a.e. convergence holds for arbitrarily small smoothing orders","New a.e. result: Bochner-Riesz converges for small order on H-type groups","Beyond known range: a.e. convergence for tiny Bochner-Riesz order on H-groups"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole theorem rests on the refined Jacobi-polynomial estimate in Theorem 8.4, namely the two bounds in (8.8) with uniform constants in the parameter range that actually occurs in the proof; if that approximation fails at its transition point or in a corner of its stated range, the second dual trace lemma loses its power and the estimate (1.13) at the trapezoid's vertex collapses.","fun_headline_variants_meta":{"raw":{"variants":["Arbitrarily small Bochner-Riesz order still gives a.e. convergence on H-type groups","A.e. convergence for Bochner-Riesz of any order on Heisenberg-type groups","H-type groups: a.e. convergence holds for arbitrarily small smoothing orders","New a.e. result: Bochner-Riesz converges for small order on H-type groups","Beyond known range: a.e. convergence for tiny Bochner-Riesz order on H-groups"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00082,"raw_usage":{"total_tokens":3522,"prompt_tokens":809,"completion_tokens":2713,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":425,"completion_tokens_details":{"reasoning_tokens":2609}},"tokens_in":425,"tokens_out":2713,"duration_ms":20369,"temperature":1.0,"reasoning_tokens":2609,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:54:49.207698+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete way to test the theorem is to evaluate the trace-lemma estimate (7.21) on a Heisenberg-type group with centre dimension $n>1$, in the middle-frequency range $\\frac34J_\\delta<j<J_\\delta$, using the explicit kernel formula (2.44) for the second-layer weight; if for a sequence $\\delta\\to0$ some test function violates the claimed bound with exponent $\\frac23$, the trace lemma is false and the theorem collapses. A more local check is to compute the two sides of (8.8) numerically for Jacobi parameters $\\alpha\\approx c(1+n)$ with $x$ near the transition point $x_{\\mathrm{tr}}$; any systematic violation would pinpoint the failing step.","supporting_citations":[{"cited_title":"Carbery, J.L","cited_arxiv_id":null,"evidence_quote":"Supplies the Euclidean almost-everywhere-convergence template and the trace-lemma approach that this paper adapts."},{"cited_title":"Gorges and D","cited_arxiv_id":null,"evidence_quote":"Proves the Heisenberg-group case and contributes the reduction to non-maximal operators and fractional integration on the group-Fourier dual."},{"cited_title":"Dunster, ‘Asymptotic approximations for the Jacobi and ultraspheri cal polynomials, and related functions’, Methods Appl","cited_arxiv_id":null,"evidence_quote":"Provides the Jacobi-polynomial asymptotics used to prove the transition-point estimate Theorem 8.4."},{"cited_title":"Boyd and T.M","cited_arxiv_id":null,"evidence_quote":"Gives the uniform asymptotic theory behind the approximation in (8.14), with a turning point and a regular singularity."},{"cited_title":"Haagerup and H","cited_arxiv_id":null,"evidence_quote":"Supplies the weighted Jacobi inequality used in the proof of Corollary 8.3."},{"cited_title":"Krasikov, ‘An upper bound on Jacobi polynomials’, J","cited_arxiv_id":null,"evidence_quote":"Gives the refined Jacobi upper bound stated as Theorem 8.2(iii) and used in the large-degree regime."},{"cited_title":"Koornwinder, A","cited_arxiv_id":null,"evidence_quote":"Provides the Jacobi Bernstein-type inequality stated as Theorem 8.2(i)."},{"cited_title":"Martini and D","cited_arxiv_id":null,"evidence_quote":"Defines the spectral-multiplier threshold $\\varsigma_+(L)$ on which the basic estimates (1.15) and Theorem 1.2 rest."},{"cited_title":"Mauceri and S","cited_arxiv_id":null,"evidence_quote":"Establishes the general stratified-group maximal Bochner–Riesz bound that Theorem 1.1 goes beyond."}],"review_version":1}