{"id":"e9f6f2b9-c31f-485c-9bda-63a766296a7e","arxiv_id":"2507.09962","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Lacunary delta-discretised spherical maximal operators are Lp bounded for 1<p<infinity and H1 to weak L1, with constants independent of delta; the multi-parameter variant is also Lp bounded.","lead":"New estimates show that maximal averages over thin spherical shells with dyadic radii are bounded on all Lp spaces, with constants that stay bounded as the shells thin out. The work connects recent discretised spherical maximal operators to classical lacunary spherical maximal theory and adds a multi-parameter version.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 5's bootstrap depends on an unproved L^2(ℓ^∞) vector-valued inequality; without it the interpolation at p=4/3 has the wrong ℓ^r space and no decay exponent is established.","rationale":"The reader correctly identified Section 5's bootstrap as the weakest point, and I agree that the final verdict should remain conditional. The one-parameter L^p claim could likely be recovered by the classical Duoandikoetxea–Rubio de Francia mechanism from the Fourier decay estimate (2.2), so the real risk is Theorem 1.4. My concern sharpens the reader's: it is not merely that the iteration is under-displayed, but that the displayed vector-valued L^2(ℓ^∞) estimate does not follow from Lemma 5.1, and replacing it with the componentwise L^2(ℓ^2) bound changes the interpolation output to the wrong ℓ^r space. I am not asserting the theorem is false; the endpoint H^1 proof is detailed, Lemma 5.1 is plausible, and a corrected bootstrap may exist. But as written, the proof of the decay exponent for p close to 1 has a gap that must be closed before Theorem 1.4 can be considered established.","tokens_in":19714,"tokens_out":31077,"duration_ms":362581,"concrete_test":"With the definitions of Section 5 fixed, verify the displayed estimate by proving or disproving the L^2(ℓ^∞) bound for A⃗_j. Concretely, in d=2 take Ψ with supp Ψ̂ ⊂ {1/2 ≤ |ξ| ≤ 2}, set f with one nonzero piece g = f*Ψ_0, and compare the left and right sides; if |g*Ψ_0*δσ_0| is not pointwise dominated by M_0(|g|), determine whether an L^2 version can be recovered. If the only available endpoint is L^2(ℓ^2), re-run the interpolation: the conclusion becomes L^{4/3}(ℓ^{4/3}), and the subsequent Littlewood–Paley step fails because ℓ^{4/3} is not the correct square-function exponent for p<2. This settles whether Section 5 supplies a valid ε(p)>0.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5 contains the only proof of Theorem 1.4. After defining the vector-valued operator A⃗_j f = {A_k^j(f*Ψ_{k-j})}, the paper asserts ||A⃗_j f||_{L^2(ℓ^∞)} ≤ ||M_j(sup_m |f*Ψ_{m-j}|)||_2 ≲ 2^{-max j_i (d-1)/2} ||f||_{L^2(ℓ^∞)}. The first inequality is not justified. Since A_k^j g = g*Ψ_{k-j}*δσ_k, the left side is sup_k |f*Ψ_{k-j}*Ψ_{k-j}*δσ_k|, whereas the right side is sup_k |(sup_m |f*Ψ_{m-j}|)*Ψ_{k-j}*δσ_k|. Moving f*Ψ_{k-j} inside the convolution with the oscillatory kernel Ψ_{k-j} is not possible pointwise; the standard domination gives only the Hardy–Littlewood maximal function of the supremum, which carries no 2^{-max j_i} decay. If the vector-valued endpoint is instead derived componentwise from Lemma 5.1, the natural endpoint is L^2(ℓ^2), not L^2(ℓ^∞). Interpolating L^1(ℓ^1) with L^2(ℓ^2) gives L^{4/3}(ℓ^{4/3}), not L^{4/3}(ℓ^2), and the required input estimate ||(∑_k |f*Ψ_{k-j}|^{4/3})^{3/4}||_{4/3} ≲ ||f||_{4/3} is the ℓ^{4/3} Littlewood–Paley inequality, which is false for p<2. Thus the displayed inequality is exactly where the positive decay exponent ε(p) for 1<p<4/3 must be proved; it is not.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies lacunary analogues of the δ-discretised spherical maximal operators introduced by Hickman and Jančar. It claims that for d≥2 and 0<δ<1/2, the lacunary δ-discretised spherical maximal operator is bounded on L^p(R^d) for all 1<p<∞ and from H^1(R^d) to L^{1,∞}(R^d), with constants uniform in δ, and that the strong multi-parameter variant is bounded on L^p for all 1<p<∞. The authors further claim that by taking δ→0 their uniform bounds recover the classical lacunary spherical maximal estimates. The proofs combine Fourier decay estimates for annuli, almost orthogonality, the Seeger–Wright atomic decomposition of H^1, Littlewood–Paley theory, and a bootstrap argument in Section 5.","tokens_in":20124,"tokens_out":40701,"duration_ms":436120,"significance":"The results, if correct, would be a natural δ-discretised counterpart of the classical lacunary spherical maximal theory, with uniform bounds that pass to the classical limit; the H^1 endpoint would extend Christ's theorem to this discretised setting, and the multi-parameter result would parallel recent work by Lee–Lee–Oh and Hickman–Zahl. The endpoint argument in Section 4 is detailed, and the estimates for Term1,1, Term1,2, Term2,1, and Term2,2 appear to close. However, the proof of the multi-parameter L^p theorem in Section 5 contains a load-bearing gap in the vector-valued interpolation step, and the final inductive bootstrap is not actually carried out. As written, Theorem 1.4 and hence the L^p part of Theorem 1.2 are not established.","major_comments":[{"comment":"The displayed estimate after “using the L2-boundedness of M_j” reads ∥A⃗_j f∥_{L2(ℓ∞)} = ∥sup_k |A^k_j(f)|∥_2 ≤ ∥M_j(sup_m |f*Ψ_{m-j}|)∥_2 ≲ 2^{-max j_i(d-1)/2}∥f∥_{L2(ℓ∞)}. This display is not type-correct as written: the left side is a norm of a sequence operator, while A^k_j(f) on the left and the scalar f on the right are not reconciled. For scalar f the left side is sup_k |f*Ψ_{k-j}*Ψ_{k-j}*δσ_k|, whereas the right side involves sup_m |f*Ψ_{m-j}| followed by a single convolution with Ψ_{k-j}*δσ_k; a pointwise supremum cannot be moved inside a convolution with an oscillatory kernel. The standard domination gives only the Hardy–Littlewood maximal function of sup_m |f*Ψ_{m-j}| and carries no 2^{-max j_i(d-1)/2} factor. If one instead derives the bound componentwise from Lemma 5.1, the natural endpoint is L^2(ℓ^2); interpolating L^1(ℓ^1) with L^2(ℓ^2) gives L^{4/3}(ℓ^{4/3}), and the passage from ℓ^{4/3} to ℓ^2 would require an ℓ^{4/3} Littlewood–Paley inequality that is false for p<2. The displayed inequality is therefore exactly the point where the positive decay exponent for p<4/3 must be proved, and it is not proved.","section":"Section 5, vector-valued bootstrap display"},{"comment":"The final paragraph of Section 5 asserts, without proof, that “Using the similar argument inductively, we get ∥M_j f∥_{L^p} ≲ 2^{-max j_i ε}∥f∥_{L^p} for all p>1.” No induction step is displayed, and in particular no argument is given for 1<p<4/3. The L^{4/3} estimate obtained by interpolation is only the base case; the claim for p>4/3 would follow by interpolating with the trivial L∞ bound, but the range 1<p<4/3 requires a new endpoint or a genuinely different mechanism. Since the proof of Theorem 1.4 consists of summing these M_j estimates over j∈Z^d with j_i≥0, the missing ε(p)>0 for all p>1 leaves Theorem 1.4 unproved even if the L2(ℓ∞) issue in the previous comment is repaired.","section":"Section 5, final induction paragraph"}],"minor_comments":[{"comment":"In the display following the derivation of \\widehat{\\chi_{C_\\delta(0,1)}}, the variable r remains in the arguments J_{d/2}(2πr(1+δ)|ξ|) and J_{d/2}(r(1-δ)|ξ|) after the r-integral has already been evaluated; these r's should be removed.","section":"Section 2, Bessel formula"},{"comment":"The one-dimensional identity (5.1) is used to produce a tensor-product decomposition over j∈Z^d with j_i≥0, but the bookkeeping for mixed indices (where some coordinates use ψ and others use Ψ) is not written out; please clarify how A^k_j is defined for such mixed vectors.","section":"Section 5, multi-scale decomposition"},{"comment":"The estimate of Term1,3 uses the L^2 boundedness of M^δ_lac, citing Theorem 1.2(1), whose proof is only supplied later in Section 5 as a corollary of Theorem 1.4; the authors should either prove this L^2 bound independently before Section 4 or explicitly state the forward dependency.","section":"Section 4, Eq. (4.7)"},{"comment":"There are numerous typos and grammatical slips: “Plancheral” should be “Plancherel”, “do not exists” should be “does not exist”, “the last but second” should be “second to last”, and “Ωk+1” should be “Ω_{κ+1}”; these should be corrected in a revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The H^1 endpoint proof in Section 4 is the most developed part of the paper and appears largely sound. The multi-parameter L^p proof in Section 5, however, is not complete: the L^2(ℓ∞) vector-valued estimate is not justified as written, and the bootstrap for 1<p<4/3 is missing. If the authors can supply a valid vector-valued estimate or replace the bootstrap with a different argument, the results are likely plausible; as it stands I cannot recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper introduces a natural lacunary analogue of the δ-discretised spherical maximal operators and a multi-parameter strong variant, with uniform-in-δ constants. That is a legitimate new question, and the H^1 → L^{1,∞} endpoint proof in Section 4 is the most careful part: the Fourier decay estimates for the annulus are worked out properly, and the use of the Seeger–Wright atomic decomposition with stopping times is detailed.\n\nThe soft spot is Section 5, the proof of the L^p bounds. The bootstrap for the multi-parameter operator depends on the displayed inequality ∥A⃗_j f∥_{L^2(ℓ^∞)} ≤ ∥M_⃗j( sup_m |f * Ψ_{m-j}| )∥_2. This domination is not justified. The left side has sup_k |(f * Ψ_{k-j}) * Ψ_{k-j} * δσ_k|; the right side feeds sup_m |f * Ψ_{m-j}| into the same oscillatory convolution. Pointwise control of f * Ψ_{k-j} by the sup does not survive convolution with the signed, cancellative kernel Ψ_{k-j} * δσ_k. So the inequality is false in general.\n\nThis is not a minor fix. Theorem 1.4 is proved only through this bootstrap, and Theorem 1.2(1) is presented as a corollary. The componentwise L^2 estimates from Lemma 5.1 give at best an L^2(ℓ^2) bound, and interpolating that with L^1(ℓ^1) lands at ℓ^{4/3}, the wrong Littlewood–Paley space. The positive decay exponent for p < 4/3 is exactly what is unproved.\n\nThe good news: the one-parameter L^p result is probably true. The uniform Fourier decay (2.2) should feed directly into the Duoandikoetxea–Rubio de Francia argument, giving L^p for all p>1 without the multi-parameter machinery. The H^1 endpoint would then follow from the L^2 bound. So the paper's core one-parameter claims are likely salvageable; the multi-parameter theorem needs either a corrected vector-valued inequality or a different approach.\n\nMinor issues: a wrong cross-reference in Section 4.1.2 (the stopping time (4.9) appears earlier), and the notational slip in the definition of A⃗_j.\n\nBottom line: this deserves serious referee time, but the referee must push on Section 5. If the authors fix that, and prove the one-parameter case directly, the paper will be solid. For now, the proof of Theorem 1.4 is incomplete.","headline":"A real gap in the Section 5 bootstrap—the L^2(ℓ^∞) domination appears false—but the one-parameter claims are likely salvageable via classical methods.","tokens_in":20652,"tokens_out":11003,"would_cite":false,"duration_ms":103448,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B30","42B20","42B35"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that the lacunary $\\delta$-discretised spherical maximal operator is bounded on $L^p(\\mathbb{R}^d)$ for all $1<p<\\infty$, uniformly in the annulus thickness $\\delta$, with an endpoint $H^1 \\to L^{1,\\infty}$ bound…","keywords":["lacunary maximal function","delta-discretised spherical maximal operator","multi-parameter maximal operator","Hardy space H^1","weak-type endpoint","Fourier decay","multi-scale square-function decomposition","bootstrapping argument"],"falsifier":"Run the displayed $L^{4/3}$ interpolation step for the vector-valued operator $\\vec{A}_{\\vec{j}}$ and iterate it to obtain the decay exponent $\\varepsilon_n(p)$ after $n$ steps; then check whether $\\inf_{p>1}\\lim_{n\\to\\infty}\\varepsilon_n(p)$ is positive. If for any $p>1$ the accumulated exponent is not bounded below by a positive constant independent of $\\vec{j}$, the summation over $\\vec{j}$ fails and the uniform $L^p$ claim is false.","tokens_in":19532,"feed_emoji":"🔵","tokens_out":12333,"duration_ms":125545,"temperature":0.7,"pith_summary":"The paper studies maximal operators formed by averaging over thin spherical shells of thickness $\\delta$, sampled at dyadically separated (lacunary) radii, and asks whether taking the supremum over those scales preserves $L^p$ norms. It asserts that the lacunary $\\delta$-discretised spherical maximal operator is bounded on $L^p$ for every $1<p<\\infty$, and from the Hardy space $H^1$ to weak $L^1$, with constants independent of $\\delta$. It further asserts the same $L^p$ boundedness for the strong multi-parameter operator, in which each coordinate dilates independently. Because the constants stay bounded as the shells shrink, letting $\\delta \\to 0^+$ recovers the classical boundedness of the lacunary spherical maximal function. A sympathetic reader should care because the uniform-in-$\\delta$ control is exactly what connects discretised spherical maximal theory to the classical lacunary estimates.","feed_headline":"Lacunary sphere averages bounded on all L^p, uniformly in δ","feed_subtitle":"Uniform bounds keep constants independent of δ, so classical lacunary spherical maximal estimates follow in the limit.","key_machinery":"The central objects are the $\\delta$-discretised spherical maximal operators $M^\\delta_{\\mathrm{lac}}f(x)=\\sup_{k\\in\\mathbb{Z}}|f*_\\delta\\sigma_k(x)|$, where $f*_\\delta\\sigma_k$ averages $f$ over the $\\delta$-neighbourhood $C^\\delta(0,2^k)$ of the sphere of radius $2^k$, and the strong multi-parameter analogue $\\mathcal{M}^\\delta_{\\mathrm{lac}}$ with coordinatewise dilations $2^{\\vec{k}}y=(2^{k_1}y_1,\\dots,2^{k_d}y_d)$. The argument is carried by two Fourier decay estimates for the averaged annulus: a $\\delta$-dependent bound of order $\\delta(1+|\\xi|)^{-(d-1)/2}$ and a $\\delta$-free bound of order $(1+|\\xi|)^{-(d+1)/2}$, which together control $\\|g*\\varphi_j*_\\delta\\sigma\\|_{L^2}$. The $H^1$ endpoint uses an atomic decomposition of the Hardy space into pieces supported near dyadic cubes, with stopping-time parameters chosen so that almost-orthogonality, exceptional-set volume, and $L^2$ estimates combine into the weak-type bound; the $L^p$ proof uses a multi-scale square-function decomposition and a bootstrap that converts an $L^2$ decay of $2^{-\\max\\{j_1,\\dots,j_d\\}(d-1)/2}$ into a positive decay for every $p>1$.","core_discovery":"The central claim is that, for $d \\ge 2$ and $0<\\delta<1/2$, the lacunary $\\delta$-discretised spherical maximal operator $M^\\delta_{\\mathrm{lac}}f(x)=\\sup_{k\\in\\mathbb{Z}}|f*_\\delta\\sigma_k(x)|$ is bounded on $L^p(\\mathbb{R}^d)$ for all $1<p<\\infty$ and from $H^1(\\mathbb{R}^d)$ to $L^{1,\\infty}(\\mathbb{R}^d)$, and that the strong multi-parameter operator $\\mathcal{M}^\\delta_{\\mathrm{lac}}$ is bounded on $L^p(\\mathbb{R}^d)$ for all $1<p<\\infty$, all with implicit constants independent of $\\delta$. The proof obtains two Fourier decay rates for the normalised Fourier transform of the $\\delta$-neighbourhood of the unit sphere, one carrying a factor of $\\delta$ and one independent of $\\delta$, and combines them with an atomic decomposition of $H^1$ and stopping-time arguments for the endpoint, and with a multi-scale square-function decomposition, an $L^2$ decay estimate, and a vector-valued interpolation bootstrap for the $L^p$ estimates. The uniformity in $\\delta$ is the point: it makes the passage to the classical lacunary spherical maximal estimates legitimate.","pith_inferences":["The load-bearing point to check is the bootstrap: the paper asserts the iteration that yields a positive decay exponent $\\varepsilon(p)>0$ for every $p>1$, but does not display the interpolation endpoints, so the claim that the range is all $p>1$ rests on an unshown induction.","A concrete check would be to compute the decay exponent explicitly from the displayed $L^{4/3}$ step and iterate the same interpolation; if the exponent tends to zero as $p$ approaches $1$, the stated all-$p$ range would fail even though each fixed $p>1$ might still be fine.","The same two-decay structure suggests a natural extension to $\\delta$-discretised lacunary averages over other hypersurfaces, provided the corresponding annulus Fourier transform obeys the same pair of bounds."],"forward_implications":["If the uniform bounds are correct, shrinking the annuli to exact spheres recovers the classical $L^p$ boundedness of the lacunary spherical maximal function and its $H^1 \\to L^{1,\\infty}$ endpoint.","The same limit argument recovers the classical strong multi-parameter lacunary spherical maximal estimates, since the constants in Theorem 1.4 are also uniform in $\\delta$.","The two Fourier decay rates for the annulus are reusable: the proof structure applies to any lacunary maximal operator built from a measure with a comparable $\\delta$-dependent and $\\delta$-free decay pair.","The endpoint proof demonstrates a template for $H^1 \\to L^{1,\\infty}$ bounds of discretised maximal operators that combines atomic decomposition with stopping times rather than a single Fourier-transform estimate."],"supporting_citations":[{"why":"Supplies the classical $L^p$ boundedness of lacunary maximal operators under Fourier decay, the result recovered in the $\\delta \\to 0^+$ limit.","marker":"[8]"},{"why":"Supplies the classical $H^1 \\to L^{1,\\infty}$ endpoint for the lacunary spherical maximal function that the uniform bound recovers.","marker":"[6]"},{"why":"Introduced the $\\delta$-discretised spherical maximal operators whose lacunary analogue is studied here.","marker":"[13]"},{"why":"Supplies the special atomic decomposition of $H^1$ used in the endpoint proof.","marker":"[24]"},{"why":"Provides the strong spherical maximal operator model and the strong maximal function bound used for the zero-frequency piece.","marker":"[16]"},{"why":"Provides the improved strong spherical maximal estimates that motivate the multi-parameter variant.","marker":"[14]"},{"why":"Supplies the special-function asymptotics used to derive the Fourier decay estimates for the annulus.","marker":"[27]"},{"why":"Supplies the level-set decomposition idea used to build the atoms in the $H^1$ decomposition.","marker":"[4]"}],"fun_headline_variants":["Lacunary sphere averages bounded on all L^p, uniform in δ","δ-independent constants for lacunary spherical maximal L^p bounds","All L^p and endpoint H^1 weak-type for lacunary δ-discretised spheres","Uniform δ bounds make classical lacunary sphere estimates follow"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that the iterative bootstrap in Section 5 actually produces a positive decay exponent $\\varepsilon(p)>0$ for every $p>1$, uniformly in the multi-scale parameter, even though the induction is asserted rather than displayed; the multi-parameter $L^p$ theorem, and the one-parameter $L^p$ bound presented as its corollary, would collapse if that exponent vanished for some $p>1$.","fun_headline_variants_meta":{"raw":{"variants":["Lacunary sphere averages bounded on all L^p, uniform in δ","δ-independent constants for lacunary spherical maximal L^p bounds","All L^p and endpoint H^1 weak-type for lacunary δ-discretised spheres","Uniform δ bounds make classical lacunary sphere estimates follow"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000875,"raw_usage":{"total_tokens":3783,"prompt_tokens":942,"completion_tokens":2841,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":558,"completion_tokens_details":{"reasoning_tokens":2758}},"tokens_in":558,"tokens_out":2841,"duration_ms":25066,"temperature":1.0,"reasoning_tokens":2758,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:45:20.182652+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the displayed $L^{4/3}$ interpolation step for the vector-valued operator $\\vec{A}_{\\vec{j}}$ and iterate it to obtain the decay exponent $\\varepsilon_n(p)$ after $n$ steps; then check whether $\\inf_{p>1}\\lim_{n\\to\\infty}\\varepsilon_n(p)$ is positive. If for any $p>1$ the accumulated exponent is not bounded below by a positive constant independent of $\\vec{j}$, the summation over $\\vec{j}$ fails and the uniform $L^p$ claim is false.","supporting_citations":[{"cited_title":"Duoandikoetxea and J","cited_arxiv_id":null,"evidence_quote":"Supplies the classical $L^p$ boundedness of lacunary maximal operators under Fourier decay, the result recovered in the $\\delta \\to 0^+$ limit."},{"cited_title":"Christ,Weak type(1, 1) bounds for rough operators, Ann","cited_arxiv_id":null,"evidence_quote":"Supplies the classical $H^1 \\to L^{1,\\infty}$ endpoint for the lacunary spherical maximal function that the uniform bound recovers."},{"cited_title":"Hickman and A","cited_arxiv_id":null,"evidence_quote":"Introduced the $\\delta$-discretised spherical maximal operators whose lacunary analogue is studied here."},{"cited_title":"Seeger and J","cited_arxiv_id":null,"evidence_quote":"Supplies the special atomic decomposition of $H^1$ used in the endpoint proof."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the strong spherical maximal operator model and the strong maximal function bound used for the zero-frequency piece."},{"cited_title":"Stein,Harmonic analysis: real-variable methods, orthogonality, and oscillatory inte- grals, Princeton University Press, Princeton, NJ., 43 (1993), xiv+695","cited_arxiv_id":null,"evidence_quote":"Supplies the special-function asymptotics used to derive the Fourier decay estimates for the annulus."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the level-set decomposition idea used to build the atoms in the $H^1$ decomposition."}],"review_version":1}