{"id":"fec36779-00b7-44f3-bc53-bff8e6c99aa9","arxiv_id":"2607.02176","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Uniform sharp bounds for the Dunkl kernel on all reduced root systems imply absolute continuity of its representing measure for multiplicities k>1/2.","lead":"The paper proves uniform upper bounds for the Dunkl kernel on any reduced root system, with the bound also valid on singular hyperplanes and for derivatives. The estimates imply that the representing measure of the Dunkl intertwining operator is absolutely continuous for multiplicities larger than one half, settling a conjecture in that range.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Induction hypothesis is not quantified over spectral parameters, so Lemma 4.5's bound on all W-translates of Φ(0) is unjustified as written.","rationale":"The reader identified Proposition 2.1(4) as the weakest assumption, but for p=1 that estimate has a constant independent of x and y, so it appears sound. The more serious soft spot is internal: Lemma 4.5 requires uniform control of Exp_k(iwλ,x0) for all w∈W, while the stated induction hypothesis controls only Exp_k(iλ,·) within one sector. This missing λ-quantifier undermines the seed of the bootstrap and is load-bearing for the proof of Theorem 2.2 and hence for Theorem 7.2. The issue is concrete and testable; it is likely fixable by running the induction simultaneously for all regular spectral parameters, since the regions a0∪a_{i+1} are independent of λ and the argument otherwise does not use special features of λ. Because the fix is straightforward and the main theorem is plausible, I do not recommend rejection, but the proof as written needs this clarification before acceptance.","tokens_in":26515,"tokens_out":40989,"duration_ms":327373,"concrete_test":"Check Lemma 4.5 against the stated hypothesis in the A_2 case: take i=1, x0∈a_2 with ⟨α1,x0⟩=⟨α2,x0⟩=a≥1, and w=s_{α1}; verify that w^{-1}x0∉S and that the fixed-λ hypothesis (4.5) gives no information on Exp_k(is_{α1}λ,x0). If a rigorous derivation of Lemma 4.5 requires a bound for Exp_k(is_{α1}λ,x0), then reformulate the induction hypothesis as: for every λ'∈a_reg, |Exp_k(iλ',y)|≤C_{λ'}|Ω_k^{-1/2}(y)| for all y∈a0∪a_{i+1}, and verify that the step i+1→i in Section 4 remains valid under this stronger hypothesis.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Theorem 2.2 (Section 4) fixes a spectral parameter λ∈a_reg and states the induction hypothesis (4.5) as a bound on Exp_k(iλ,x0) for x0 in a0∪a_{i+1} inside one fixed sector S. Lemma 4.5 then asserts ∥Φ(0)∥ ≤ C|Ω_I^{-1/2}(x0)|, where Φ(0) has entries Exp_k(iwλ,x0) for every w∈W. This does not follow from (4.5). W-equivariance gives Exp_k(iwλ,x0)=Exp_k(iλ,w^{-1}x0), and w^{-1}x0 is generally outside S: for A_2, taking i=1, x0 on a_2 with ⟨α1,x0⟩=⟨α2,x0⟩=a≥1, and w=s_{α1}, yields w^{-1}x0 with simple-root coordinates (-a,2a), outside a+. Thus the stated single-λ hypothesis gives no control on these entries. The missing ingredient is a simultaneous bound for all regular spectral parameters wλ at x0 itself, i.e., the induction must be run uniformly for all λ'∈a_reg with constants C_{λ'}. Without this, the Gronwall bootstrap in Lemma 4.8/4.10 cannot be seeded: a rough a^{μ_I} bound on ∥Φ(0)∥ would persist through the iteration, since the initial-data term is never reduced. This is a genuine gap in the written proof, though it appears fixable by quantifying the induction over λ∈a_reg.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves uniform upper bounds for the Dunkl kernel Exp_k(iλ,x) with fixed regular spectral parameter λ and Re(k)≥0, improving the rough polynomial growth estimate of de Jeu. The proof inductively moves from regular arguments to singular ones by constructing first-order systems (3.4) associated with parabolic subgroups, transforming them via a Levinson-type Q(t), and applying Gronwall-type bootstrap estimates. The same framework is extended to derivatives of the kernel and to complex spectral parameters λ=λ_1+iλ_2 with λ_1 in the closed positive chamber. As applications, the author derives L^p integrability of the Dunkl kernel and proves that, for k>1/2 and regular λ, the representing measure μ_k^λ is absolutely continuous with respect to Lebesgue measure, with density in L^2, and with higher regularity for larger k.","tokens_in":26950,"tokens_out":10981,"duration_ms":101492,"significance":"If correct, the main theorem is a substantial advance: it gives sharp, spatially uniform bounds for the Dunkl kernel and its derivatives for arbitrary reduced root systems, generalizing Clerc's estimates for Cartan motion groups and settling the absolute-continuity conjecture of Rösler–de Jeu at least for k>1/2. The method is original and largely self-contained: explicit first-order systems, precise estimates for the trigonometric-integral remainders, and bootstrap arguments replace the geometric tools used by Clerc. The paper also gives a clean roadmap for the geometric comparison and, in the appendix, supplies the deferred technical estimates. The main weakness is a gap in the inductive control of the W-translates appearing in the initial condition for Φ; this is load-bearing but appears repairable by strengthening the induction hypothesis.","major_comments":[{"comment":"The bound ∥Φ(0)∥≤C|Ω_I^{-1/2}(x0)| stated in Lemma 4.5 does not follow from the induction hypothesis (4.5) as written. Hypothesis (4.5) controls only Exp_k(iλ,x0) for x0∈a0∪a_{i+1} inside the fixed sector S, while Φ(0) contains entries Exp_k(iwλ,x0) for every w∈W. Using W-equivariance, Exp_k(iwλ,x0)=Exp_k(iλ,w^{-1}x0), and w^{-1}x0 is generally outside S: for A_2, taking x0∈a_2 and w=s_{α1} gives simple-root coordinates (-a,2a), which are not in a0∪a_{i+1}. This is not a cosmetic issue: Lemma 4.8 and Lemma 4.10 seed their Gronwall-type bootstrap with the good bound on ∥Φ(0)∥ (resp. ∥Z(T)∥). If only the rough polynomial bound of Lemma 4.7 is available, the term a^{μ_I} from the initial data persists through the iteration and the final |Ω_I^{-1/2}| estimate cannot be reached. The gap appears fixable by strengthening the induction to run simultaneously on all finitely many sectors (or uniformly over the finite W-orbit of λ), with constants depending on λ; the same repair is needed for Lemma 5.2 in the derivative case and for Lemma 6.3 in the complex-spectral-parameter case.","section":"§4, Lemma 4.5 and Eq. (4.5), (4.9); also §5, Lemma 5.2 and §6, Lemma 6.3"}],"minor_comments":[{"comment":"The phrase 'successively lowering the exponent' is terse; the number of iterations is finite and bounded in terms of μ_I and m_I independently of a, but this should be stated explicitly so the uniformity of the constants is fully transparent.","section":"§4, Lemmas 4.8 and 4.10"},{"comment":"The inequality |Ω_k^{-1/2}(x0)|≤|ω_I^{-1/2}(x0)Ω_I^{-1/2}(x0)| should be written with a constant C (or justified via Lemma 4.1), since Ω_k and ω_IΩ_I are comparable but not equal.","section":"§4, Eq. (4.12)"},{"comment":"The extension of Lemma 5.1 to general polynomials p is described only by saying the proof still works; a short sentence explaining which estimates are supplied by the induction hypothesis on lower-degree derivatives would improve readability.","section":"§5, Appendix 9.1"},{"comment":"Several cross-references call lemmas 'Theorem' (for example, 'Theorem 4.5' in the proof of Lemma 4.6 and 'Theorem 4.7' in the proof of Lemma 4.10); these should be corrected for consistency.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of the journal and the main result is significant. The gap concerning W-translates in the induction is real but appears to be fixable by a finite simultaneous induction over sectors or spectral parameters; I therefore recommend major revision rather than rejection. The external input from [dJ93] used to seed the bootstrap is standard and acceptable, but the authors should make the strengthened induction hypothesis explicit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proves sharp uniform upper bounds for the Dunkl kernel for arbitrary reduced root systems and derives the Rösler–de Jeu absolute continuity conjecture for k>1/2. The core ideas are genuinely new: the parabolic sector induction, the first-order system with the Q-transformation, and the bootstrap that sharpens rough polynomial growth into the correct Ω_k^{-1/2} decay. The application to absolute continuity is clean and the paper is well-written, self-contained, and properly cites the prior literature. This is serious work that deserves refereeing.\n\nThat said, there is a real gap in the written proof, exactly where the stress-test note points. Lemma 4.5 claims a bound on ∥Φ(0)∥ involving all W-translates of the spectral parameter, but the induction hypothesis (4.5) is stated only for a single fixed λ and points x0 inside one fixed sector S. The step from (4.5) to Lemma 4.5 uses W-equivariance to replace Exp_k(iwλ,x0) by Exp_k(iλ,w^{-1}x0), but w^{-1}x0 need not lie in S. The A2 example with s_{α1} maps (a,a) to (-a,2a), so the induction hypothesis for S gives no control on that entry. The same issue underlies the earlier claim that it suffices to prove the bound on a+: for x in another chamber one needs the bound for a different spectral parameter wλ on a+. The constants in the theorem are allowed to depend on λ, so this is fixable by running the induction uniformly over the finite orbit Wλ (or over all regular λ) and taking the worst constant. With that quantification, Lemma 4.5 would follow and the bootstrap in Lemmas 4.8 and 4.10 would be properly seeded. The fix is straightforward and I do not see any load-bearing obstruction beyond it.\n\nOther soft spots are minor: a few bootstrap steps are stated tersely (\"successively lowering the exponent\"), and Lemma 4.8's iteration deserves a few more lines, but the appendix fills in the harder estimates. The comparison |ω_I^{-1/2}Ω_I^{-1/2}| ≤ C|Ω_k^{-1/2}| is valid on the sector because ⟨α,x⟩ is bounded away from zero for α∉R_I. No circularity, no fitted parameters, no suspicious citation pattern.\n\nWho should read this: anyone working in Dunkl harmonic analysis, Bessel functions, or spherical functions of Cartan motion groups. It deserves a serious referee. I would recommend sending to a journal and asking for a revision that quantifies the induction over the W-orbit; after that fix, the result is likely correct.","headline":"Strong results with a genuine but fixable gap in the induction's quantification over the W-orbit.","tokens_in":27329,"tokens_out":10974,"would_cite":true,"duration_ms":96587,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["33C52","33C67"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes uniform spatial bounds for the Dunkl kernel at regular spectral parameters, and derives absolute continuity of the representing measure for k > 1/2 as a consequence.","keywords":["Dunkl kernel","Dunkl operators","uniform estimates","root systems","Levinson theorem","absolute continuity","representing measure","Bessel functions"],"falsifier":"On the root system $A_2$ with $k=0.75$ and regular $\\lambda$, compute $|\\operatorname{Exp}_k(i\\lambda,x)|$ along the ray $\\langle\\alpha_1,x\\rangle=\\langle\\alpha_2,x\\rangle=t$ for large $t$ and compare it with $C\\prod_{\\alpha\\in R_+}(1+|\\langle\\alpha,x\\rangle|)^{-1.5}$; any polynomial growth of the ratio with $t$ refutes Theorem 2.2.","tokens_in":26320,"feed_emoji":"📐","tokens_out":9407,"duration_ms":77073,"temperature":0.7,"pith_summary":"This paper proves that, for any reduced root system and multiplicities with $\\operatorname{Re} k\\ge 0$, the Dunkl kernel with regular imaginary spectral parameter is uniformly controlled by the weight $|\\Omega_k^{-1/2}(x)|$: there is a constant $C_{R,k,\\lambda}$ such that $|\\operatorname{Exp}_k(i\\lambda,x)|\\le C|\\Omega_k^{-1/2}(x)|$ for all $x\\in \\mathfrak{a}$. The bound remains valid as $x$ approaches the singular hyperplanes, where earlier estimates only gave polynomial growth. The main consequence is that the compactly supported probability measure representing Dunkl's intertwining operator is absolutely continuous with respect to Lebesgue measure whenever $k>1/2$ and $\\lambda$ is regular, with density in $L^2$. This settles, for $k>1/2$, a conjecture previously known only in rank-one, $A_2$, and symmetric-group cases.","feed_headline":"Uniform Dunkl bounds settle a 2002 conjecture for k > 1/2","feed_subtitle":"A weighted decay estimate for the Dunkl kernel forces its representing measure to have an L² density for regular spectral parameters.","key_machinery":"The load-bearing object is the first-order system $\\partial_H \\Phi = A\\Phi$ whose entries are $F_w(x)=\\omega_I^{1/2}(x)e^{-i\\langle x,w\\lambda\\rangle}\\operatorname{Exp}_k(iw\\lambda,x)$ along a ray $x_0+tH_i$ into a singular stratum, with $\\omega_I$ the product of $|\\langle\\alpha,x\\rangle|^{2k_\\alpha}$ over roots not vanishing on $H_i$. The matrix $A$ selects reflections $s_\\alpha w$ and oscillates, so it is only conditionally integrable. The decisive step is a change of variable $Z=(1+Q)^{-1}\\Phi$ with $Q=-\\int_t^\\infty A(s)\\,ds$; the new coefficient $(1+Q)^{-1}AQ$ decays like $(a+t)^{-2}$, and a Gronwall-type bootstrap lowers the polynomial growth exponent step by step until it crosses the threshold forced by $\\Omega_I^{-1/2}$, yielding the desired bound. For non-purely-imaginary spectral parameters the same construction is modified with $Q$ solving $Q'=[\\Gamma,Q]+A$ to keep the exponential factors under control.","core_discovery":"The central discovery is Theorem 2.2: for a regular spectral parameter $\\lambda\\in \\mathfrak{a}_{\\mathrm{reg}}$ and $\\operatorname{Re} k\\ge 0$ there is a constant $C_{R,k,\\lambda}$ such that $|\\operatorname{Exp}_k(i\\lambda,x)|\\le C|\\Omega_k^{-1/2}(x)|$ for every $x\\in \\mathfrak{a}$, where $\\Omega_k(x)=\\prod_{\\alpha\\in R_+}(1+|\\langle\\alpha,x\\rangle|)^{2k_\\alpha}$. This is a uniform bound that stays meaningful at the walls of the Weyl chamber, where the old polynomial-growth estimate could not distinguish directions. The proof reaches it by an induction over strata of a Weyl chamber, following the geometric strategy used for spherical functions, but replaces geometric input with a first-order system built directly from the Dunkl eigenvalue equation and a bootstrap that successively lowers polynomial exponents until the $\\Omega_k^{-1/2}$ weight appears. Theorem 7.2 then follows: for $k>1/2$ and $\\lambda\\in \\mathfrak{a}_{\\mathrm{reg}}$ the representing measure $\\mu_k^\\lambda$ has density in $L^2(\\mathfrak{a})$.","pith_inferences":["The threshold $k>1/2$ is probably not the true boundary; the proof uses $L^2$ membership as its finishing step, so the same scheme may establish absolute continuity for all $k>0$ through $L^p$ arguments, or reveal a genuine singularity at exactly $k=1/2$. This is an editorial inference, not a claim of the paper.","The sector-bootstrap recipe—stratify a Weyl chamber, build a conditionally integrable first-order system along rays into singular strata, transform by $Q$ to achieve $L^1$ coefficients, then lower growth exponents—looks transferable to other Dunkl-type eigenfunctions and to related hypergeometric eigenfunctions outside the conical regime; the paper only gestures at these comparisons.","In the complex case $k=1$, the density should be recoverable explicitly and compared with the known piecewise-polynomial orbital-integral densities; the Sobolev-regularity statement in the paper gives a quantitative starting point for that comparison."],"forward_implications":["For $k>1/2$ and regular $\\lambda$, the representing measure $\\mu_k^\\lambda$ is absolutely continuous with density in $L^2(\\mathfrak{a})$, resolving the 2002 conjecture in this range.","For $k>1$ the density is continuous, and for $k>\\ell+(n+1)/2$ it is of class $C^\\ell$; the same Sobolev argument shows the Fourier transform of the Dunkl kernel is $C^\\ell$ for $\\operatorname{Re} k$ above the same threshold.","The Dunkl kernel $\\operatorname{Exp}_k(i\\lambda,\\cdot)$ lies in $L^p(\\mathfrak{a})$ whenever $\\operatorname{Re} k>1/p$, improving the earlier polynomial-growth integrability statement uniformly up to the walls.","The derivative bounds give uniform control on $p(\\partial)\\operatorname{Exp}_k(i\\lambda,\\cdot)$ with the same $\\Omega_k^{-1/2}$ weight, previously available only in rank one and in geometric Cartan-motion cases.","Averaging the kernel recovers the sharp uniform bounds for spherical functions on Cartan motion groups and for classical one-variable Bessel functions."],"supporting_citations":[{"why":"Supplies the polynomial-growth bound on the Dunkl kernel that seeds the initial estimates in the bootstrap.","marker":"[dJ93, Lemma 3.5]"},{"why":"Gives the first-order ODE system for the Dunkl kernel on regular cones and poses the absolute-continuity conjecture.","marker":"[RdJ02, Corollary 3]"},{"why":"Provides the geometric prototype of the induction from regular to singular directions that the paper adapts to Dunkl operators.","marker":"[Cle87]"},{"why":"Levinson theorem whose proof structure is modified to transform the first-order system into one with absolutely integrable coefficients.","marker":"[Eas89, Thm. 1.11.1]"},{"why":"Constructs the matrix $Q$ solving $Q'=[\\Gamma,Q]+A$ used for non-purely-imaginary spectral parameters.","marker":"[BL15, Thm. 4.26]"},{"why":"Supplies the integral representation of the Dunkl kernel with a compactly supported probability measure, the object of the absolute-continuity application.","marker":"[Rö99]"},{"why":"Fourier inversion for tempered distributions converts the $L^2$ membership of the kernel into absolute continuity of the representing measure.","marker":"[Tre67, Ch. 24]"}],"fun_headline_variants":["Uniform Dunkl kernel bounds prove L^2 density for k > 1/2","Dunkl kernel bounds settle 2002 conjecture for k > 1/2","Sharp Dunkl estimates yield absolute continuity for k > 1/2","Uniform bounds on Dunkl kernel resolve old conjecture","Dunkl kernel uniform bounds imply measure density for k > 1/2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument inherits the rough polynomial-growth estimate of the Dunkl kernel from earlier work, and the whole sector induction would have no seed if that estimate's constants secretly grew with $x$ near the singular hyperplanes.","fun_headline_variants_meta":{"raw":{"variants":["Uniform Dunkl kernel bounds prove L^2 density for k > 1/2","Dunkl kernel bounds settle 2002 conjecture for k > 1/2","Sharp Dunkl estimates yield absolute continuity for k > 1/2","Uniform bounds on Dunkl kernel resolve old conjecture","Dunkl kernel uniform bounds imply measure density for k > 1/2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000737,"raw_usage":{"total_tokens":3265,"prompt_tokens":888,"completion_tokens":2377,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":2278}},"tokens_in":504,"tokens_out":2377,"duration_ms":13238,"temperature":1.0,"reasoning_tokens":2278,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:35:48.540719+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the root system $A_2$ with $k=0.75$ and regular $\\lambda$, compute $|\\operatorname{Exp}_k(i\\lambda,x)|$ along the ray $\\langle\\alpha_1,x\\rangle=\\langle\\alpha_2,x\\rangle=t$ for large $t$ and compare it with $C\\prod_{\\alpha\\in R_+}(1+|\\langle\\alpha,x\\rangle|)^{-1.5}$; any polynomial growth of the ratio with $t$ refutes Theorem 2.2.","supporting_citations":[],"review_version":2}