{"id":"c90f9db6-7614-4042-a126-9f7987466f48","arxiv_id":"1908.06458","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every H1 function on a λ-Dirichlet group is almost everywhere the pointwise limit of its first Riesz means of any positive order, via a new weak-type maximal inequality.","lead":"Defant and Schoolmann prove that H1 functions on Dirichlet groups are almost everywhere the pointwise limit of their first Riesz means, for every frequency. The key new tool is a weak-type (1,∞) Hardy-Littlewood maximal operator adapted to these groups, with applications to the infinite-dimensional torus and general Dirichlet series.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.10 as stated is false: (26) defines a supremum over all intervals, not a pointwise maximal operator; the proof requires intervals anchored at 0, so the weak-type bound behind Theorem 2.1 and Corollary 2.6 is unproved.","rationale":"The paper's central construction is otherwise coherent: the Riesz kernel estimates (Lemmas 3.3, 3.4) are plausible, the transference Lemma 1.4 is sound, and the applications follow formally once Theorem 2.1 is available. But Theorem 2.10 is not a minor typo: as stated it asserts weak (1,1) boundedness of a non-pointwise maximal operator that is infinite a.e. for natural L1 functions. The proof of Lemma 3.1 accidentally proves the centered version. Since Theorem 2.1 is the sole route to Corollary 2.6, the paper's main claim is currently unproved as written. The concern is internal inconsistency, not disagreement with consensus. A local repair (centered maximal operator) appears available, so a conditional acceptance is more accurate than outright rejection. I do not see a comparable problem in the self-cited framework; the reader's weakest assumption, reliance on [4]/[19], is a fair external risk but not the decisive point here.","tokens_in":30041,"tokens_out":15927,"duration_ms":175227,"concrete_test":"Counterexample to (26)/Theorem 2.10: on G=T with βT(t)=e^{-it}, set δ=1/(2H) and f_H(z)=H on the arc |arg z|<δ, 0 elsewhere, so ‖f_H‖1=1. For each ω choose t0 with ωe^{-it0}=1; then (1/|I|)∫_I |(f_H)_ω|dt = H for I=(t0−δ,t0+δ), hence M(f_H)(ω)≥H for every ω. For α=H/2, m({M(f_H)>α})=1, so the weak-type inequality in Theorem 2.10 forces its constant to be ≥H/2 for arbitrarily large H, a contradiction. To test the repair, replace (26) by centered intervals (or intervals containing 0) and re-run Lemma 3.1 and Proposition 3.2; if the maximal inequality and Corollary 2.6 follow, the fix is purely local.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The load-bearing step is Theorem 2.10, used to prove Theorem 2.1 and hence Corollary 2.6. The operator defined in (26), M(f)(ω)=sup_{I⊂R} |I|^{-1}∫_I |fω(t)|dt, is not the classical Hardy-Littlewood maximal operator: it takes a supremum over all intervals in the time line, with no condition that the interval is anchored at the point where the function is evaluated. On G=T, β(t)=e^{-it}, this makes M(f)(ω) essentially the global supremum of arc averages of |f|, which can be infinite a.e. even for L1 functions with an integrable spike. In Lemma 3.1 the proof selects, for each t∈Ω_ω(α), an interval I_t containing t with 1/|I_t|∫_{I_t}|fω|>α; under (26) the interval provided by M^A(f)(ωβ(t))>α is t+J_t with J_t⊂[-A,A], and t+J_t need not contain t (it contains t only if 0∈J_t). Thus the Vitali covering step does not cover Ω_ω(α). The argument becomes correct if (26) is replaced by the centered maximal operator sup_{r>0} (2r)^{-1}∫_{-r}^r |fω(t)|dt; the later use in Proposition 3.2 via [9, Thm 2.1.10] only needs this centered version. As printed, however, the stated Theorem 2.10 is false, and the proof of Theorem 2.1 — and therefore Corollary 2.6 — is unsupported.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a theory of Riesz means for Fourier series of functions in Hardy spaces H^λ_p(G) on λ-Dirichlet groups and applies it to general Dirichlet series. The main claim is that every f in H^λ_1(G) is almost everywhere equal to the pointwise limit of its first (λ,k)-Riesz means for any k>0, and correspondingly that almost all vertical limits of an H_1(λ)-Dirichlet series are (λ,k)-Riesz summable almost everywhere on the imaginary axis. The technical engine is a claimed weak-type (1,∞) Hardy-Littlewood maximal operator on Dirichlet groups, Theorem 2.10. From this the authors derive the maximal inequality for Riesz means (Theorem 2.1), pointwise convergence results (Corollaries 2.2, 2.3, 2.6), norm convergence (Theorem 2.13), and an identification of H^λ_∞(G) with uniformly almost periodic holomorphic functions on the right half-plane (Theorem 2.16). The paper also contains negative results showing that second Riesz means fail for the frequencies (n) and (log n).","tokens_in":30405,"tokens_out":11448,"duration_ms":121948,"significance":"If the main results hold, they constitute a substantial contribution: they extend classical a.e. convergence theory for H_1(T) to a general frequency framework, give a positive answer for logarithmic means on the infinite-dimensional torus where Cesàro means fail, and provide a clean transfer between Fourier series on Dirichlet groups and vertical limits of Dirichlet series. The transference lemma (Lemma 1.4) and the reduction of a.e. summability to a maximal inequality are conceptually attractive, and the applications to H_1(T^∞) and to uniformly almost periodic functions are natural and interesting. However, the central maximal operator as defined is not the classical Hardy-Littlewood operator and, as stated, Theorem 2.10 is false. Because Theorem 2.1 and its corollaries rest directly on that theorem, the current proof of the main claims is unsupported. The intended argument appears repairable by replacing the operator with its centered version, so the paper has clear potential, but it requires substantial correction before it can be accepted.","major_comments":[{"comment":"The operator M defined in (26) is not the Hardy-Littlewood maximal operator: the supremum is taken over all intervals I⊂R without requiring that the interval is anchored at the point where the function is evaluated. As stated, Theorem 2.10 is false. For the (n)-Dirichlet group (T,β_T), let f be the indicator of a small arc A⊂T of length ε. For every ω, the function f_ω is a periodic indicator of an interval of length ε, and since the supremum in (26) ranges over all intervals, one may choose I equal to one full period of the support, giving M(f)(ω)=1 for all ω. Then m({M(f)>1/2})=1, which cannot be bounded by C/(1/2)·‖f‖_1=C·ε/2 uniformly as ε→0. The defect is visible in the proof of Lemma 3.1: from M^A(f)(ωβ(t))>α one only obtains an interval I⊂[-A,A] whose average of |f_{ωβ(t)}| exceeds α; after the change of variables this is an average over t+I, which need not contain t, so the intervals used in the Vitali covering argument need not cover the set Ω_ω(α). The argument becomes correct if (26) and (30) are replaced by the centered maximal operator sup_{r>0}(2r)^{-1}∫_{-r}^r |f_ω(t)|dt and its finite-A truncation; then the interval obtained is centered at t and the proof of Lemma 3.1 works. This is a load-bearing correction because Theorem 2.1 and Corollary 2.6 are deduced from Theorem 2.10.","section":"§2.3, Eq. (26); §3.1, Lemma 3.1"},{"comment":"The proof of Proposition 3.7 uses the unrestricted supremum in a second place: it asserts that for all intervals I and u>0, (1/|I|)∫_I |(f_ω∗P_u)(t)|dt ≤ ∫ P_u(a)(1/|I|)∫_I |f_ω(t-a)|dt da ≤ M(f)(ω). With the corrected centered definition of M, the inner average is over the interval I-a, which is not centered at 0, so the displayed pointwise bound no longer follows. The proposition itself is presumably still true, but the proof must be revised, for instance by using the comparability of centered and uncentered Hardy-Littlewood maximal operators, once (26) is changed. Since Proposition 3.7 is used for part (20) of Corollary 2.2 and hence for Corollary 2.6(25), this revision is necessary for the full statement of the main results.","section":"§3.3, Proposition 3.7"}],"minor_comments":[{"comment":"Equation (16) defines S^{λ,k}_x(D)(s) with the factor e^{λ_n s}; to be consistent with D(s)=∑ a_n e^{-λ_n s} and with the earlier definition of second Riesz means, the factor should be e^{-λ_n s}.","section":"§1.7, Eq. (16)"},{"comment":"The displayed formula for Re^{λ,k}_N(D)(s) contains an extraneous factor a_n, writing a_n(1-n/N)a_n n^{-s}; it should be ∑_{n<N} a_n(1-n/N)n^{-s}.","section":"§1.7, display before (17)"},{"comment":"The paper relies heavily on the authors' preprints [4] and [19] for the entire H^λ_p framework, the vertical restriction lemma, and the Bohr-Cahen formula (58). These are listed as 'preprint 2019' and 'to appear'; the authors should state their current status and, if necessary, make the relevant statements available for verification.","section":"References [4] and [19]"}],"recommendation":"major_revision","confidential_remarks":"The false definition of the maximal operator is a serious, load-bearing flaw, but it is also clearly local: the intended centered Hardy-Littlewood maximal operator is standard, and the proof of Lemma 3.1 becomes correct under that change. The remaining uses of M in Proposition 3.7 and elsewhere require adjustment but not a new strategy. I therefore recommend major_revision rather than rejection. I would also ask the editor to ensure that the foundational results imported from [4] and [19] are available and that the sign typo in equation (16) and the duplicated coefficient in (17) are corrected in the revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper's central claim is unproved. Theorem 2.10, which anchors the weak-type inequality for Riesz means, defines its maximal operator as a supremum over all intervals of the restriction f_ω, with no anchor at any point. That operator is not the Hardy-Littlewood maximal operator; it can be infinite almost everywhere for an L1 function with tall narrow spikes. The proof of Lemma 3.1 selects, for each t in the covering set, an interval containing t, but the operator's definition does not provide such an interval. The Vitali covering step therefore collapses.\n\nThe fix is evident: replace (26) with the centered maximal operator sup_r (2r)^{-1}∫_{-r}^r |f_ω(t)|dt, which is what the proof of Proposition 3.2 actually uses (via the convolution-at-0 estimate from Grafakos, Theorem 2.1.10). With that change, Lemma 3.1 works, because an average over (-r,r) of f_{ωβ(t)} corresponds to an interval (t-r,t+r) in the t-line, which does contain t. So I would bet the main theorem is true, but the version in front of us does not prove it.\n\nThe paper is not a throwaway. The Riesz-means framework for Dirichlet groups is well organized, the reduction of the Riesz maximal inequality to a Hardy-Littlewood bound is elegant, and the applications to H1(T^∞) and to vertical limits of Dirichlet series are genuinely new. The reliance on the authors' earlier preprints [4] and [19] is substantial, but those are background tools, and the key transfer lemma (Lemma 1.4) is proved in the paper. Other soft spots are minor: a few \"in preparation\" references, some typos, and the section on second Riesz means is fine but not deep.\n\nWho is this for? Harmonic analysts working on general Dirichlet series and H1 summability. The ideas deserve serious attention, but the manuscript needs correction before the main claims are accepted. As is, I would not accept it; I would invite a revision with the maximal operator fixed and Lemma 3.1 rewritten. That is a small change mathematically, but it is load-bearing. Send it out to a referee who can check the fix, and if it holds, publish.","headline":"The paper's central maximal inequality is built on a misdefined Hardy-Littlewood operator; the result is likely repairable but the main theorem is unproved as written.","tokens_in":30966,"tokens_out":6612,"would_cite":false,"duration_ms":64633,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["43A17","30H10","30B50"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every H1 function on a Dirichlet group is recovered almost everywhere by its first Riesz means; equivalently, almost every vertical limit of an H1-Dirichlet series is (λ,k)-Riesz summable on the imaginary axis.","keywords":["Fourier series on groups","general Dirichlet series","vertical limits","Riesz summability","Hardy spaces","Dirichlet groups","Hardy-Littlewood maximal operator"],"falsifier":"One would refute the main theorem by finding a frequency λ, k>0, and f∈H^λ_1(G) such that the set of ω∈G where the first Riesz means fail to converge to f has positive Haar measure, or where sup_x |$R^{{λ,k}}$_x(f)(ω)| is infinite on a set of positive measure. For the ordinary case λ=(log n), a concrete search would check whether any non-convergent H1(T) Fourier series, embedded in H1(T∞), has a vertical limit whose logarithmic means diverge on a positive-measure set of t.","tokens_in":29821,"feed_emoji":"📈","tokens_out":10103,"duration_ms":92581,"temperature":0.7,"pith_summary":"The paper proves that integrable Hardy-space functions on Dirichlet groups, despite the failure of ordinary Fourier convergence, are almost everywhere recovered by their first Riesz means of any positive order k. Equivalently, for every frequency λ, almost all vertical limits of an H1-Dirichlet series are (λ,k)-Riesz summable almost everywhere on the imaginary axis. The engine is a new weak-type (1,∞) Hardy–Littlewood maximal operator on Dirichlet groups, which controls the supremum of the Riesz means and turns a density argument for polynomials into an almost-everywhere statement for all H1 functions. A reader should care because p=1 is exactly the boundary case where Carleson–Hunt convergence fails, and Riesz summation restores pointwise recovery in a way adapted to the exponents λ_n.","feed_headline":"Riesz means recover H1 Dirichlet series almost everywhere","feed_subtitle":"A weak-type maximal inequality makes almost every vertical limit of an H1 Dirichlet series summable on the boundary line.","key_machinery":"The load-bearing object is a Hardy–Littlewood maximal operator defined on a Dirichlet group (G,β): Mf(ω)=sup_I |I|^{-1}∫_I |f_ω(t)| dt, where f_ω(t)=f(ωβ(t)) is the vertical restriction of f along the dense flow β:R→G. Theorem 2.10 proves M is weak type (1,1) and strong type (p,p) for p>1 by reducing it, via a Vitali covering argument and Fubini, to the one-dimensional maximal theorem. Proposition 3.2 then bounds the Riesz maximal operator pointwise by a constant multiple of Mf(ω), so all almost-everywhere convergence of Riesz means follows from the classical maximal theorem plus density of Dirichlet polynomials. A second mechanism is Lemma 1.4, the transference lemma that converts a.e. convergence on G into a.e. convergence of vertical limits of Dirichlet series on R, and the Bohr–Cahen formula from [19] that controls the abscissa of uniform Riesz summability in the H∞ application.","core_discovery":"The central discovery is that first Riesz means are the right summation method for H1 Hardy spaces on Dirichlet groups. The paper shows (Theorem 2.1) that the maximal operator $R^{{λ,k}}$_{max} f(ω)=sup_{x>0}|∑_{λ_n<x} \\hat f(h_{λ_n})(1−λ_n/x)^k h_{λ_n}(ω)| is bounded from H^λ_1(G) into weak L1(G) and from H^λ_p(G) into Lp(G) for p>1, for every frequency λ and k>0. From this maximal inequality the authors deduce almost-everywhere convergence of the first Riesz means to f (Corollary 2.2), and via the Bohr transform the same statement for almost all vertical limits of H1-Dirichlet series on the imaginary axis (Corollary 2.6). Along the way they establish an H1-norm approximation theorem and an isometric identification of H^λ_∞(G) with bounded holomorphic uniformly almost periodic functions on the right half-plane.","pith_inferences":["The new maximal operator M likely applies to other a.e. convergence questions on Dirichlet groups, since it gives a differentiation theorem for integrable functions along the flow β (Corollary 2.11) and a Besicovitch-norm interpretation of the vertical restrictions f_ω.","The argument suggests that any summation kernel whose L1-control can be expressed through the one-dimensional maximal function will satisfy the same weak-type bound; this could be tested for weighted or multi-parameter Riesz means.","If the identification in Theorem 2.16 holds for arbitrary frequencies, the Bohr–Cahen formula may be usable to characterize which λ admit boundary extension of bounded holomorphic functions on Re>0 without Landau's condition.","The failure of second Riesz means for λ=(log n) hints that no summability method intermediate between first Riesz means and ordinary convergence can work uniformly across all frequencies; describing the class of λ for which eλ-summability holds would be a natural next step."],"forward_implications":["For the infinite-dimensional torus T∞, every f∈H1(T∞) is almost everywhere the pointwise limit of its logarithmic Riesz means for any k>0; the corresponding Cesàro means fail in general.","For any H1 Dirichlet series D, almost every vertical limit D_ω is (λ,k)-Riesz summable at every point u+it with u≥0 and almost every t, including the boundary line Re=0.","The H1 norm approximation R^{λ,k}_x(f)→f holds for every f∈H^λ_1(G) and k>0.","The space H^λ_∞(G) is isometrically and coefficient-preservingly the same as the bounded holomorphic functions on Re>0 that are uniformly almost periodic on vertical lines and have Bohr spectrum contained in λ.","First Riesz means of any positive order work; second Riesz means do not in general, for λ=(n) and λ=(log n), so the result cannot be strengthened to eλ-summability."],"supporting_citations":[{"why":"Builds the H^λ_p-theory of general Dirichlet series: the isometric Bohr transform identification, the a.e. existence of vertical restrictions f_ω, and the density of Dirichlet polynomials, on which the whole transference argument rests.","marker":"[4]"},{"why":"Supplies the definition of (λ,k)-Riesz means together with the consistency theorems and the integral identities used in the maximal estimates and in Lemmas 3.8 and 3.9.","marker":"[10]"},{"why":"Gives the Bohr–Cahen formula for the abscissa of uniform Riesz summability, used to prove the H∞ identification in Theorem 2.16.","marker":"[19]"},{"why":"Supplies the weak L^1 space machinery, Vitali covering, Marcinkiewicz interpolation, and the one-sided maximal function bound used in the proof of the maximal operator theorem.","marker":"[9]"},{"why":"Provides the definitions and standard facts on uniformly almost periodic functions and Bohr coefficients needed for the H∞ application.","marker":"[2]"},{"why":"Establishes the reflexive-case convergence theorem for Dirichlet series on the infinite-dimensional torus, which the p=1 result extends and contrasts with.","marker":"[12]"}],"fun_headline_variants":["Riesz means converge a.e. for H1 Dirichlet series","Maximal inequality tames Riesz means on Dirichlet groups","Almost all vertical limits are Riesz-summable","Weak-type bound yields a.e. Riesz convergence","New maximal operator proves Riesz summability a.e."],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument presupposes the validity of the authors' earlier H^λ_p-theory of general Dirichlet series: every H1 Dirichlet series must be representable, isometrically and coefficient-preserving, as an integrable function on a compact group with a dense one-parameter flow, so that vertical restrictions and Fubini arguments apply.","fun_headline_variants_meta":{"raw":{"variants":["Riesz means converge a.e. for H1 Dirichlet series","Maximal inequality tames Riesz means on Dirichlet groups","Almost all vertical limits are Riesz-summable","Weak-type bound yields a.e. Riesz convergence","New maximal operator proves Riesz summability a.e."]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000567,"raw_usage":{"total_tokens":2665,"prompt_tokens":906,"completion_tokens":1759,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":522,"completion_tokens_details":{"reasoning_tokens":1678}},"tokens_in":522,"tokens_out":1759,"duration_ms":11547,"temperature":1.0,"reasoning_tokens":1678,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:45:44.528082+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One would refute the main theorem by finding a frequency λ, k>0, and f∈H^λ_1(G) such that the set of ω∈G where the first Riesz means fail to converge to f has positive Haar measure, or where sup_x |$R^{{λ,k}}$_x(f)(ω)| is infinite on a set of positive measure. For the ordinary case λ=(log n), a concrete search would check whether any non-convergent H1(T) Fourier series, embedded in H1(T∞), has a vertical limit whose logarithmic means diverge on a positive-measure set of t.","supporting_citations":[{"cited_title":"Defant and I","cited_arxiv_id":null,"evidence_quote":"Builds the H^λ_p-theory of general Dirichlet series: the isometric Bohr transform identification, the a.e. existence of vertical restrictions f_ω, and the density of Dirichlet polynomials, on which the whole transference argument rests."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the definition of (λ,k)-Riesz means together with the consistency theorems and the integral identities used in the maximal estimates and in Lemmas 3.8 and 3.9."},{"cited_title":"Schoolmann: On Bohr’s theorem for general Dirichlet series , to appear in Math","cited_arxiv_id":null,"evidence_quote":"Gives the Bohr–Cahen formula for the abscissa of uniform Riesz summability, used to prove the H∞ identification in Theorem 2.16."},{"cited_title":"Grafakos: Classical Fourier analysis , Graduate Texts in Mathematics 249 (2014)","cited_arxiv_id":null,"evidence_quote":"Supplies the weak L^1 space machinery, Vitali covering, Marcinkiewicz interpolation, and the one-sided maximal function bound used in the proof of the maximal operator theorem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the definitions and standard facts on uniformly almost periodic functions and Bohr coefficients needed for the H∞ application."},{"cited_title":"Hedenmalm and E","cited_arxiv_id":null,"evidence_quote":"Establishes the reflexive-case convergence theorem for Dirichlet series on the infinite-dimensional torus, which the p=1 result extends and contrasts with."}],"review_version":1}