{"id":"705549fa-fba1-49fc-8116-78c3eddaf3ea","arxiv_id":"2501.16262","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For two-step stratified Lie groups with degenerate brackets satisfying Assumptions A and B, sharp-order p-specific spectral multiplier bounds hold for p up to an explicit range, extending nondegenerate results.","lead":"This paper proves sharp-order L^p bounds for spectral multipliers on a new class of two-step stratified Lie groups whose bracket structure is degenerate. It is a mathematical advance in harmonic analysis, with consequences for Bochner-Riesz summability and sub-Riemannian PDE.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Main theorem's p-range formula is internally inconsistent: the displayed p_{\\bar d1,d2} lacks a factor 2 that the final derivation and Example 2.2 both use.","rationale":"The reader's weakest_assumption, Assumption B, is a real restriction but it is explicitly built into the class of groups studied and verified for the examples; it is not an internal inconsistency. The most load-bearing internal issue is the mismatch between the displayed formula for p_{\\bar d1,d2} in Theorem 1.1 and the value actually derived and used in the proof. The final computation after (7.16) algebraically yields the doubled numerator, and Example 2.2's numerical evaluation 2 - 8/(N+7) is exactly that formula. The displayed theorem therefore states a smaller p-range than the argument establishes; correcting it strengthens the statement. This is not a rejection-level gap, but it is precisely the kind of error that blocks acceptance because the theorem as printed is internally inconsistent. The proposed concrete check is a direct symbolic re-derivation of the final algebra and a numerical evaluation at N=2. My overall assessment agrees with the reader's conditional verdict, so I would not change it.","tokens_in":33690,"tokens_out":15610,"duration_ms":136791,"concrete_test":"Perform the algebra from Eq. (7.16) exactly: substitute 1/q = 1/2 - 1/p' and theta_p = pST(d2)'/p' with pST(d2)' = 2(d2+1)/(d2-1) into (7.16), and check whether the resulting upper bound for p has numerator 2\\bar d1(d2-1) + 2d2(d2+1) or \\bar d1(d2-1) + 2d2(d2+1). Then evaluate both candidate formulas at (\\bar d1,d2)=(2,3) and compare with the value 2 - 8/(N+7) from Example 2.2; only the doubled formula gives 1 at N=1 and 10/9 at N=2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 defines p_{\\bar d1,d2} = [\\bar d1(d2-1) + 2d2(d2+1)] / [\\bar d1(d2-1) + 3d2^2 + 1]. The proof's last step, after Eq. (7.16), derives instead p <= [2\\bar d1(d2-1) + 2d2(d2+1)] / [\\bar d1(d2-1) + 3d2^2 + 1], and Example 2.2 evaluates the threshold at (\\bar d1,d2)=(2N,3) as 2 - 8/(N+7), which is exactly the doubled-numerator expression. The two disagree: for N=2, the displayed formula gives 8/9, while the derived and example value is 10/9. Thus the main theorem as printed states a strictly smaller p-range than the argument actually proves; the definition and the example cannot both be correct. Because the p-range is the quantitative content of the claimed p-specific multiplier theorem, this is a load-bearing internal inconsistency. The fix is to add the missing factor 2 in the first numerator term of Theorem 1.1 and to correct the subscript p_{d1,d2} to p_{\\bar d1,d2}. This strengthens the statement, so it does not by itself invalidate the method, but the paper cannot be accepted with the theorem in its current form.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves new L^p spectral multiplier estimates for sub-Laplacians on two-step stratified Lie groups whose group structure is degenerate, under two structural assumptions (Assumptions A and B) and a dimensional condition \\bar d_1 := 2(r_1+...+r_N) \\geq d_2-1. For 1 < p \\leq p_{\\bar d_1,d_2} it obtains boundedness of F(L) with the sharp p-specific Sobolev regularity s > d|1/p-1/2|, together with the corresponding Bochner-Riesz summability order. The proof combines a dyadic reduction, restriction-type estimates for cap-localized joint functional calculus pieces, Plancherel estimates for convolution kernels, and a two-stage localization of the kernels in the first and second layers. Applications are given to glued copies of the free two-step group N_{3,2} and to Heisenberg-Reiter groups.","tokens_in":1962,"tokens_out":2069,"duration_ms":126073,"significance":"If the proof is correct, this is a substantial advance: it extends p-specific spectral multiplier theorems with Euclidean sharp regularity to a class of two-step groups where the matrices J_\\mu are allowed to have nontrivial kernel, beyond the previous Metivier and Heisenberg-type cases. The cap decomposition on the unit sphere in the center is a genuine technical novelty, and the paper is carefully structured, with detailed support-localization propositions and explicit exponent counts. The result is conditional on several external ingredients, notably [Mar15, Lemma 10] and restriction-type estimates from [Nie24b], but these are existing theorems rather than ad-hoc assumptions. The main theorem, after correction of the p-threshold formula, is quantitatively stronger than the version printed in Theorem 1.1.","major_comments":[{"comment":"The displayed formula for p_{\\bar d_1,d_2} in Theorem 1.1 is internally inconsistent with the derivation at the end of Section 7.4 and with Example 2.2. The theorem defines p_{\\bar d_1,d_2} = [\\bar d_1(d_2-1)+2d_2(d_2+1)]/[\\bar d_1(d_2-1)+3d_2^2+1], but the proof after Eq. (7.16) derives p \\leq [2\\bar d_1(d_2-1)+2d_2(d_2+1)]/[\\bar d_1(d_2-1)+3d_2^2+1], and Example 2.2 evaluates the threshold at (\\bar d_1,d_2)=(2N,3) as 2-8/(N+7), which agrees with the doubled-numerator formula. For N=2 the displayed formula gives 8/9, whereas the derived and example value is 10/9. Since the p-range is the quantitative content of the theorem, this must be fixed; after adding the missing factor 2 the statement is strengthened and matches the proof.","section":"Theorem 1.1 and Section 7.4, Eq. (7.16)"},{"comment":"The proof uses the identification pST(d_1,d_2)=pST(d_2), justified by the sentence “our assumptions imply in particular d_1 \\geq d_2.” The displayed condition (1.2) only gives \\bar d_1 \\geq d_2-1, and since d_1 = \\bar d_1 + r_0 with r_0 \\geq 0, it does not by itself imply d_1 \\geq d_2. This point is load-bearing because the final exponent count and the definition of \\theta_p depend on which of pST(d_1) and pST(d_2) is smaller. Please add a proof or a precise reference for the assertion d_1 \\geq d_2 under Assumptions A and (1.2); if it is not always true, the proof of Theorem 4.1 and the final threshold need to be revised.","section":"Section 7.4, text before Eq. (7.13)"}],"minor_comments":[{"comment":"The condition “1 \\leq p \\leq p_{d_1,d_2}” should read “1 \\leq p \\leq p_{\\bar d_1,d_2}”, using the corrected subscript introduced in the definition of the threshold.","section":"Theorem 1.1"},{"comment":"The text before Proposition 5.2 says the cap size is chosen as \\delta = R/R_\\ell, while the proof of Lemma 5.3 uses \\delta = R_\\ell/R. For \\ell \\leq \\iota the first choice gives a cap size at least 1, which is inconsistent with the use of small caps; the proof indicates that the intended value is \\delta = R_\\ell/R, so the earlier sentence should be corrected or clarified.","section":"Section 5.2 and proof of Lemma 5.3"},{"comment":"The sentence “Since the sets supp \\chi_j have only bounded overlap” should refer to the sets supp \\zeta_j, not \\chi_j.","section":"Proof of Lemma 6.2"},{"comment":"The notation f_r(x,u)=f(rx,r^2u) and the accompanying reduction for arbitrary \\chi \\in C_c^\\infty(R_+) are very compressed; a short explanation of why the substitution preserves the hypotheses of [Mar15, Lemma 10] would improve readability.","section":"Remark 6.4"},{"comment":"In the Heisenberg-Reiter example, the sentence “2r_n := rank((J^{H_{N,d_2}}_\\mu)^2 P_n) = rank((J^{H_{1,d_2}}_\\mu)^2)” is correct, but the following line “=1” could be misunderstood because the rank is 2, not 1; please check the displayed computation of 2(r_1+...+r_N).","section":"Section 2.4, Example 2.3"}],"recommendation":"major_revision","confidential_remarks":"The main technical content appears sound and the central issue is a fixable inconsistency in the stated p-range, which the proof actually strengthens. I recommend major revision rather than rejection. The paper relies on [Nie24b] and [Nie25], which are preprints or recently accepted; the editor may wish to ensure these are accessible to referees. The scope restriction imposed by Assumption B is significant, and it would help if the final version gave a clearer general discussion of which degenerate groups satisfy it beyond the two examples."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing to know: this is a real advance, not a repackaging. Niedorf proves sharp p-specific spectral multiplier estimates on a degenerate class of two-step stratified groups where only non-sharp Mikhlin-Hörmander bounds were known, with concrete examples including glued copies of N3,2 and Heisenberg-Reiter groups. The cap decomposition on the center sphere is a genuine new device. The proof is long but structured: restriction-type estimates, support localization, Plancherel bounds, and the final exponent count are all present. I did not find a fatal gap.\n\nThe soft spots are real but manageable. Assumption B is a genuine restriction: it excludes degenerate groups whose kernel carries a nonzero bracket, and the paper is upfront about that. The proof leans heavily on [Mar15, Lemma 10] and on the author's own restriction and cluster estimates from [Nie24b] and [Nie25]. I could not verify every cited estimate, but those are independent prior theorems, not circular dependencies.\n\nThe main problem is an internal inconsistency in the stated theorem. Theorem 1.1 defines p_{\\bar d1,d2} = [\\bar d1(d2-1) + 2d2(d2+1)] / [\\bar d1(d2-1) + 3d2^2 + 1]. The final derivation around (7.16) gives instead p <= [2\\bar d1(d2-1) + 2d2(d2+1)] / [\\bar d1(d2-1) + 3d2^2 + 1], and Example 2.2 uses that doubled numerator to obtain 2 - 8/(N+7). The two disagree: for N=2 the displayed formula gives 8/9 while the derived and example value is 10/9. Since the p-range is the quantitative content of the theorem, the paper cannot be accepted with the formula as printed. The fix is clear and it strengthens the statement, but the definition and the example cannot both be correct. Also, the subscript in the theorem should be p_{\\bar d1,d2}, not p_{d1,d2}.\n\nBottom line: the central argument appears sound, Assumption B narrows the class, and the formula error is theorem-level but easily repaired. This deserves a serious referee. I would want the referee to check the applications of [Mar15, Lemma 10] and [Nie24b, Theorem 4.1] exactly as stated, and to re-verify the exponent count in Section 7.4 after the formula correction.","headline":"Real new result for degenerate two-step groups, but the main theorem's p-range formula has a missing factor 2 and must be corrected before acceptance.","tokens_in":34498,"tokens_out":3003,"would_cite":true,"duration_ms":26481,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B15","22E25","22E30","43A85"],"pacs":[],"model":"deepseek-v4-flash","headline":"On two-step stratified Lie groups with degenerate group structure satisfying Assumptions A and B and \\bar d_1 \\ge d_2 - 1, the paper proves sharp p-specific L^p spectral multiplier estimates and Bochner–Riesz summability with regularity…","keywords":["two-step stratified Lie group","sub-Laplacian","spectral multiplier","Bochner-Riesz means","restriction type estimate","sub-Riemannian geometry","Heisenberg-Reiter group","degenerate group structure"],"falsifier":"Exhibit a two-step stratified Lie group that satisfies Assumption A and \\bar d_1 \\ge d_2 - 1 but has two vectors x,x' \\in \\ker J_{\\mu_0} with [x,x'] \\neq 0 for some \\mu_0 \\neq 0. On such a group Assumption B fails and the proof's inequality (6.5), which needs |[x,x']| \\le C R_\\ell R after projecting out the kernel components, cannot hold; this gives a concrete test of whether the theorem's conclusion survives without Assumption B.","tokens_in":33445,"feed_emoji":"📐","tokens_out":13403,"duration_ms":107215,"temperature":0.7,"pith_summary":"On a class of two-step stratified Lie groups where the group law is allowed to be degenerate, the paper proves that every sub-Laplacian L admits L^p spectral multiplier estimates with the sharp Euclidean regularity s > d|1/p-1/2|. The result is new because earlier p-specific estimates of this type required the non-degeneracy (Métivier) condition, whereas here the bracket matrices J_\\mu may have a kernel of positive dimension. The class is fixed by two assumptions: a simultaneous block decomposition of the matrices J_\\mu with one nonzero eigenvalue per block (Assumption A), and the condition that any two kernel vectors of the same J_\\mu bracket to zero (Assumption B). The main theorem yields these estimates for 1 < p \\le p_{\\bar d_1,d_2}, where \\bar d_1 = 2(r_1+\\dots+r_N) \\ge d_2-1, and as a corollary gives uniform L^p bounds for Bochner–Riesz means (1-tL)_+^\\delta for \\delta > d(1/p-1/2)-1/2. The examples include copies of the free two-step nilpotent group N_{3,2} glued along their centers and Heisenberg–Reiter groups satisfying the dimensional condition.","feed_headline":"Degenerate Lie groups get sharp L^p multiplier bounds","feed_subtitle":"Two-step stratified groups with degenerate structure reach Euclidean regularity via center-sphere caps.","key_machinery":"The machinery centres on the family of skew-symmetric matrices J_\\mu = \\mu_1 J_1 + \\dots + \\mu_{d_2} J_{d_2} that encodes the group law in exponential coordinates, together with the sub-Laplacian L = -($X_1^{2}$+\\dots+X_{d_1}^2). Assumption A gives a simultaneous spectral decomposition -J_\\$mu^{2}$ = \\sum_n (b_n^\\mu)^2 P_n^\\mu with a single nonzero eigenvalue on each block; Assumption B makes the kernel direction commute internally. The paper's new technical device is a refined joint spectral decomposition of L and the center operator U = (-($U_1^{2}$+\\dots+U_{d_2}^2))^{1/2} into caps on the unit sphere of the center, with cap size \\delta = R_\\ell/R linked to the dyadic scales. The caps fix a reference direction \\mu_j and allow the kernel of the dyadic operator F(L)\\chi(2^\\ell U)\\zeta_j(U) to be localized on the first layer at scale R_\\ell and on the second layer at scale R_\\ell R; the latter localization is exactly where Assumption B enters. The proof is completed by restriction type estimates (interpolating with the Stein–Tomas exponent p_{\\mathrm{ST}}(d_2)) and weighted Plancherel estimates for the center variable.","core_discovery":"The paper's central claim is Theorem 1.1: under Assumptions A and B and \\bar d_1 := 2(r_1+\\dots+r_N) \\ge d_2 - 1, every bounded Borel function F with finite local Sobolev norm \\|F\\|_{$L^{2}$_{s,\\mathrm{sloc}}} for s > d(1/p-1/2) gives a bounded operator F(L) on L^p(G), with \\|F(L)\\|_{p\\to p} \\le C_{p,s}\\|F\\|_{$L^{2}$_{s,\\mathrm{sloc}}}, for 1 < p \\le p_{\\bar d_1,d_2}; the endpoint p=1 is covered by the Bochner–Riesz statement. The admissible range is the explicit number p_{\\bar d_1,d_2} that the proof derives from condition (7.16); on the glued N_{3,2} example it equals 2 - 8/(N+7), so the range approaches (1,2) as the number of copies grows. Consequently the Bochner–Riesz means (1-tL)_+^\\delta are uniformly bounded on L^p(G) for \\delta > d(1/p-1/2) - 1/2. This extends the p-specific multiplier theory beyond the Métivier (non-degenerate) setting; the paper notes it recovers weaker versions of the earlier Métivier-group results when Assumption A holds there.","pith_inferences":["The cap decomposition of the center sphere, with cap size tied to the ratio of dyadic scales, is a transferable device: the same idea could localize kernels for other sub-elliptic operators whose central variables form a multi-dimensional sphere or torus; the paper only develops it for two-step stratified groups.","If Assumption B were dropped, the second-layer localization |u| \\lesssim R_\\ell R would fail at inequality (6.5); a natural test case is any group satisfying Assumption A with dim ker J_\\mu \\ge 2 and a nonzero bracket between kernel vectors, for which the present proof gives no conclusion.","Since p_{\\bar d_1,d_2} tends to 2 as \\bar d_1 grows, the theorem predicts that for very many blocks the degenerate group behaves essentially like the Euclidean case for L^p multipliers; this large-\\bar d_1 regime is not explicitly discussed in the paper."],"forward_implications":["On groups covered by Theorem 1.1, Bochner–Riesz means (1-tL)_+^\\delta are uniformly bounded on L^p for \\delta > d(1/p-1/2)-1/2, the same order as the Bochner–Riesz conjecture.","Copies of N_{3,2} glued along their centers satisfy the assumptions, with p = 2 - 8/(N+7) for N \\ge 2; the range approaches the full interval (1,2) as the number of copies increases.","Heisenberg–Reiter groups H_{N,d_2} satisfy the assumptions whenever 2N \\ge d_2 - 1, so the sharp estimates hold on these previously studied degenerate groups.","The regularity threshold s > d(1/p-1/2) is sharp by known lower bounds, so within its range the result cannot be improved in the regularity order.","For Métivier groups that also satisfy Assumption A, the theorem recovers parts of the earlier p-specific multiplier results, now as a special case of the degenerate setting."],"supporting_citations":[{"why":"Defines the degenerate class under Assumption A and supplies the second-layer weighted Plancherel estimate (its Lemma 10) used to control |u|-moments of dyadic kernels.","marker":"[Mar15]"},{"why":"Provides the simultaneous spectral decomposition of J_\\mu (Lemma 5) and the Plancherel kernel formula (Corollary 8) used in the cap-localized kernel analysis.","marker":"[MM14]"},{"why":"Supplies the restriction type estimates and spectral cluster estimate for twisted Laplacians on general two-step groups used in Theorem 3.1.","marker":"[Nie24b]"},{"why":"The Métivier-group p-specific multiplier result whose dyadic reduction and finishing steps the proof adapts for the degenerate case.","marker":"[Nie25]"},{"why":"Provides Theorem 4.1 (its Corollary 6.2) reducing p-specific multiplier estimates to dyadic spectral multiplier bounds with exponential decay.","marker":"[Nie24]"},{"why":"Gives the Mikhlin–Hörmander threshold s>Q/2 used to dismiss the non-dyadic error term in Section 7.","marker":"[Chr91]"},{"why":"Gives the general spectral multiplier theorem for stratified groups used for the same error term alongside [Chr91].","marker":"[MM90]"},{"why":"Establishes sharpness of the threshold s>d|1/p-1/2|, justifying that the paper's regularity order cannot be lowered.","marker":"[MMNG23]"},{"why":"Supplies the sub-elliptic weighted estimate for Hermite operators used to convert first-layer localization into rapid decay in Lemma 5.3.","marker":"[CO16]"},{"why":"Provides the Laguerre function norms and support estimates controlling the kernel contributions in Lemmas 3.4 and 5.3.","marker":"[Tha93]"}],"fun_headline_variants":["Degenerate Lie groups get sharp L^p bounds via center caps","Sharp multipliers on degenerate two-step Lie groups","Caps on the sphere unlock Euclidean multiplier bounds","Degenerate Lie groups reach Euclidean regularity in L^p","Beyond Métivier: sharp L^p bounds via center-sphere caps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is Assumption B: any two vectors in the kernel of the same matrix J_{\\mu_0} (for \\mu_0 \\neq 0) must have zero bracket, [x,x']=0; this is not a consequence of the block-decomposition Assumption A and is used at the step (6.5) to obtain the second-layer localization |u| \\lesssim R_\\ell R.","fun_headline_variants_meta":{"raw":{"variants":["Degenerate Lie groups get sharp L^p bounds via center caps","Sharp multipliers on degenerate two-step Lie groups","Caps on the sphere unlock Euclidean multiplier bounds","Degenerate Lie groups reach Euclidean regularity in L^p","Beyond Métivier: sharp L^p bounds via center-sphere caps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000258,"raw_usage":{"total_tokens":1596,"prompt_tokens":975,"completion_tokens":621,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":539}},"tokens_in":591,"tokens_out":621,"duration_ms":5783,"temperature":1.0,"reasoning_tokens":539,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:36:54.363193+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a two-step stratified Lie group that satisfies Assumption A and \\bar d_1 \\ge d_2 - 1 but has two vectors x,x' \\in \\ker J_{\\mu_0} with [x,x'] \\neq 0 for some \\mu_0 \\neq 0. On such a group Assumption B fails and the proof's inequality (6.5), which needs |[x,x']| \\le C R_\\ell R after projecting out the kernel components, cannot hold; this gives a concrete test of whether the theorem's conclusion survives without Assumption B.","supporting_citations":[],"review_version":1}