{"id":"48645d8c-dcf5-467a-b565-62fd2ebaf560","arxiv_id":"2411.16029","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On product cones over closed manifolds with conjugate radius larger than pi, the Schrödinger and half-wave propagators satisfy global pointwise dispersive estimates with the Euclidean decay rate times an angular weight.","lead":"This paper proves pointwise decay estimates for the Schrödinger and half-wave propagators on product cones, under the geometric condition that the cone's cross-section has no conjugate points within distance pi. The result shows the classical Euclidean decay rate |t|^{-n/2} survives up to a weight, and marks pi as the critical conjugate-radius threshold.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.3's restriction to d=d_h is load-bearing: non-distance sheets in D(y1,y2) are dismissed as 'simpler' without a uniform treatment, and a phase singularity or a bounded-z gap would break the parametrix bound.","rationale":"The reader's weakest_assumption correctly identifies the non-distance geodesic sheets and the compressed treatment in Section 4.2. I have no disagreement with that identification. The paper's central claim depends on the global parametrix of Corollary 3.6 being valid uniformly over all d in D(y1,y2), and the proof only details the distance sheet. The missing uniform treatment of non-distance sheets, especially the phase singularities and the bounded-z window where d ~ z^{-1/2}, is a genuine soft spot. However, the gap is plausibly repairable by standard microlocal cut-offs and a careful analysis of the finite distance spectrum, so the correct verdict remains CONDITIONAL rather than ACCEPT or REJECT. The proposed test on a flat torus with a very short circle would either expose a counterexample to the 'simpler' assertion or confirm that the uniform estimates hold, thus settling whether the concern actually lands.","tokens_in":48764,"tokens_out":44661,"duration_ms":431775,"concrete_test":"Take Y = S^1_a x S^1_b with a << 1 << b, so inj(Y)=a/2. Write the four shortest non-distance sheets in D(y1,y2), e.g., d_{1,0}(y1,y2)=|y1-y2+(a,0)|, and choose a pair with d_h(y1,y2) << a so that d_{1,0} ~ a. For z ~ a^{-2}, we have d_{1,0} ~ z^{-1/2}, placing the sheet exactly in the regime that Lemma 4.3's Case 2 requires d >= C1 z^{-1/2}; verify whether (4.13) holds uniformly in z,y1,y2 by direct evaluation of the oscillatory integral with phase d_{1,0} xi and symbol b(rho d_{1,0}). Also check the behavior as y1-y2 approaches (-a,0), where d_{1,0} is not differentiable. If the bound fails or a singularity appears, the 'same, in fact simpler' claim in Section 4.2 is false and the theorem is unproved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central estimate Proposition 4.2 reduces everything to Lemma 4.3, which is proved only for d(y1,y2)=d_h(y1,y2). Section 4.2 then asserts that all other sheets d in D(y1,y2) are 'the same, in fact simpler' because d is lower bounded by inj(Y)>0. This is the load-bearing step. The lower bound alone does not justify the assertion: Lemma 4.3's Case 2 requires d >= C1 z^{-1/2}, and for z in the range 1 << z <~ (C1/inj(Y))^2 a non-distance d can be smaller than z^{-1/2}; no separate argument is supplied for that bounded-z window. Equally important, Corollary 3.6 treats every d in D(y1,y2) as a smooth phase function in the oscillatory sum (3.31), but non-distance sheets are only locally smooth and have conical singularities where the sheet degenerates (e.g., on a flat torus, d_{1,0}(y1,y2)=|y1-y2+(1,0)| is not differentiable at y1-y2=(-1,0)). Proposition 3.2's proof only establishes finiteness of fibers of exp|_{B}, not that these phase singularities can be avoided or absorbed into the remainder R_N with uniform constants. If a non-distance sheet's phase singularity interacts with the stationary-phase analysis in Lemma 4.3, the bound (4.13) loses uniformity and Theorem 1.1's estimate would fail. The same sheet-wise decomposition is also used for the boundary cancellation in I_{GD} via Proposition 4.6, so the gap sits at the center of the argument. The footnote in Lemma 4.3 admitting an omitted boundary justification at rho=infinity adds a further, more standard technical gap, but the non-distance sheet issue is the primary soft spot.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies pointwise dispersive estimates for the Schrödinger propagator e^{itH} and the half-wave propagator e^{it√H} on a product cone X=C(Y) with metric g=dr^2+r^2h, where H=Δ_g+V_0(y)r^{-2}. The main result, Theorem 1.1, asserts that if the conjugate radius of the closed manifold Y satisfies R_Conj(Y)>π and the operator P=Δ_h+V_0+(n-2)^2/4 is strictly positive, then the Schwartz kernel of e^{itH} obeys the bound (1.8), with an explicit dependence on the normalized variable r_1r_2/(2|t|). The proof combines an exact representation of the propagator kernel via Bessel functions (Proposition 2.1), a Hadamard-type parametrix for cos(s√P) and the Poisson-wave propagator on Y (Section 3), and a delicate stationary-phase analysis of the resulting oscillatory integrals (Section 4). The paper also derives weighted L^1→L^∞ estimates (Corollary 1.4), L^{q'}→L^q estimates (Theorem 1.6), and Besov-space decay estimates for the half-wave propagator (Theorem 1.11) via Littlewood-Paley theory and heat-kernel bounds. A new threshold phenomenon at R_Conj=π is claimed, separating the positive result from known counterexamples on spheres of radius less than 1.","tokens_in":49109,"tokens_out":9988,"duration_ms":92700,"significance":"If the proof is completed, the paper gives the first pointwise dispersive estimates for Schrödinger and wave equations on general product cones with a sharp geometric threshold (R_Conj>π), generalizing flat-cone results (Ford, Blair-Ford-Marzuola, Zhang) and matching the sphere counterexamples (Taira). The approach is largely parameter-free: the kernel representation in Proposition 2.1 is exact, and the parametrices in Section 3 involve no fitted constants. The Littlewood-Paley and heat-kernel ingredients are standard and carefully connected to the dispersive estimates. The threshold claim is falsifiable and plausible in view of the rescaled geodesic flow structure. However, the current manuscript leaves several load-bearing technical points unproved, most notably the treatment of non-distance sheets in the parametrix sum and the corresponding stationary-phase and boundary-cancellation arguments; these gaps must be fixed before the main theorem can be considered established.","major_comments":[{"comment":"The proof of Lemma 4.3 is carried out only for d(y1,y2)=d_h(y1,y2). The claim that non-distance sheets d∈D(y1,y2) are 'the same, in fact simpler' because d is lower bounded by inj(Y)>0 is not substantiated. In Case 2, the argument requires d ≥ C1 z^{-1/2} to ensure that the β0-neighborhood {|s-d| ≤ (zd)^{-1}} stays away from s=0 and to justify the bounds in (4.24)-(4.26). For a non-distance sheet with d ≥ inj(Y), the inequality d ≥ C1 z^{-1/2} fails for the bounded-z window 1 ≪ z < (C1/inj(Y))^2; the paper supplies no separate estimate there. Since Lemma 4.3 is the only estimate behind Proposition 4.2 and hence Theorem 1.1, this is a load-bearing gap.","section":"Section 4.2, Lemma 4.3"},{"comment":"The parametrix (3.31) represents cos(s√P) as a sum over d∈D(y1,y2) and treats each d as a smooth phase function on Y×Y. Corollary 3.6 inherits this from Proposition 3.4, where each sheet of the Lagrangian L± is parametrized by φ_d = d(y1,y2)1·ξ. However, for a geodesic loop counted with multiplicity, the function d is only locally smooth and has conical singularities where different geodesics merge; e.g., on a flat 2-torus, d_{1,0}(y1,y2)=|y1-y2+(1,0)| is not differentiable on the cut locus y1-y2=(-1,0). Proposition 3.2 proves only that the fibers of exp_{y0}|_B are finite and uniformly bounded, not that the individual sheets can be chosen smooth globally or that the singularity set can be excluded from the stationary-phase and boundary-cancellation arguments. In particular, Proposition 4.6 uses the jet matching (3.37) sheet by sheet; if a sheet is only piecewise smooth, the integration-by-parts identities (4.43)-(4.44) and the cancellation in I_GD may produce boundary terms at the singularities that are not controlled uniformly in y1,y2. An explicit treatment of the non-distance sheets is therefore required.","section":"Section 3.2, Corollary 3.6"},{"comment":"The footnote at the end of the proof of Lemma 4.3 concedes that the integration by parts in dρ near ρ=+∞ is not justified and says 'we omit the details'. This is not a cosmetic point: the estimates (4.21) and (4.28) rely on integrating by parts in ρ to gain ρ^{-N} and on dropping the boundary term at infinity, and the same device is reused in the IGD estimates (4.46)-(4.59). Without a rigorous dyadic localization near infinity (or an alternative justification to interpret these as oscillatory integrals with admissible cutoffs), the uniform bound (4.13) is not established. The gap is routine in nature but must be filled, as the argument is load-bearing for Proposition 4.2.","section":"Section 4.2, proof of Lemma 4.3"}],"minor_comments":[{"comment":"The abstract uses 'conjugate radius ε' while the main text uses R_Conj; please unify the notation.","section":"Abstract / Section 1"},{"comment":"In the displayed estimate after equation (5.4), the inequality d(z1,j,z0)+d(z0,z2,j) ≥ 1/2(d(z1,j,z0)+d(z0,z2,j))^2 is dimensionally inconsistent; it should be replaced by an inequality such as (d1+d2)^2 ≥ d^2(z1,j,z2,j) together with e^{-(d1^2+d2^2)/c} ≤ e^{-(d1+d2)^2/(2c)}.","section":"Section 5, proof of Proposition 5.1"},{"comment":"The sign convention in the operator in Remark 1.3 ('−∆ + V0(y)r^{-2}') differs from that in (1.1) where H = ∆_g + V0(y)r^{-2}; please clarify the convention.","section":"Section 1, Remark 1.3"},{"comment":"In (4.31), after 2m integrations by parts, the denominator should contain ν^{2m} consistently in both integrals; the current display shows the same notation after the second integral, but the derivation in the proof would be clearer if the powers were tracked explicitly.","section":"Section 4.2, Lemma 4.4"}],"recommendation":"major_revision","confidential_remarks":"The reader's report and the skeptic's analysis identify the same core gap: the treatment of non-distance sheets in the parametrix and in the stationary-phase/boundary-cancellation arguments. I concur that this is load-bearing and not merely a presentation issue. The positive claims are plausible and the threshold π is an attractive, falsifiable finding, but the proof as written is not yet complete. The manuscript would benefit from a detailed treatment of the non-distance sheets, including a precise statement of their local smoothness, a resolution of the conical singularities, and the corresponding modification of Lemma 4.3 and Proposition 4.6. The related work by Taira (arXiv:2503.21527) suggests the threshold claim is timely and worth publishing once the technical gaps are closed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a serious advance, but it has a gap that needs a referee's heel. The main theorem—pointwise dispersive estimates for Schrödinger and wave on product cones with R_Conj > π—is new and significant: it generalizes the flat-cone results to all product cones above the π threshold, and the threshold itself is a clean geometric discovery. The proof is mostly careful. The exact kernel representation in Proposition 2.1 is rigorous, and the boundary cancellation between the cosine and Poisson-wave parametrices is a nice mechanism. No fitted parameters, no circularity.\n\nThe soft spot is exactly where the reader put it: the treatment of non-distance sheets in the distance spectrum. Section 4.2 says 'the same, in fact simpler' because such d are bounded below by inj(Y)>0. That lower bound does not by itself justify the claim for all large z. In Lemma 4.3, Case 2 needs d ≥ C1 z^{-1/2} to make the dyadic decomposition work. For a non-distance sheet, d ≥ inj(Y) gives that only when z ≥ (C1/inj(Y))^2; in the window 1 << z << (C1/inj(Y))^2, the argument as written does not apply. This window is a fixed, possibly large constant depending on Y, so it cannot be waved off as a harmless low-frequency region without an additional estimate.\n\nThe second concern, about phase singularities of non-distance sheets, is real but I think milder than the stress-test suggests. On a flat torus the branches d_{1,0} are smooth via the covering; the universal-cover singularity at (-1,0) maps to the diagonal, where the branch is actually smooth. Still, the paper never shows how the local parametrization of Proposition 3.4 globalizes to all of Y×Y with uniform remainder. That globalization is not automatic and should be written down.\n\nThe footnote in Lemma 4.3 about the boundary at ρ=∞ is a minor omission; standard oscillatory-integral reasoning fixes it.\n\nNone of this makes me think the theorem is false. The structure is right, and the gaps look addressable. The paper deserves a serious referee, not a desk reject. The referee should insist on a complete treatment of the non-distance sheets and a global partition-of-unity argument in Section 3.2. I'd bring it to reading group and cite it if the gaps close.","headline":"A substantial, plausible result with a real but likely fixable gap in the non-distance sheet analysis; deserves a serious referee.","tokens_in":49669,"tokens_out":9360,"would_cite":true,"duration_ms":86050,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q41","35L05","58J40"],"pacs":[],"model":"deepseek-v4-flash","headline":"On product cones, if the cross-section has no conjugate point pairs within distance $\\pi$, the Schr\\\"odinger and half-wave kernels obey Euclidean-type pointwise decay.","keywords":["pointwise dispersive estimates","product cones","Schrodinger propagator","half-wave propagator","conjugate radius","Hadamard parametrix","inverse-square potential","conical singular spaces"],"falsifier":"Compute the Schr\\\"odinger kernel (2.1) on the cone over $Y=S^{n-1}_\\sigma$ with $\\sigma<1$ under the same positivity assumption on $P$, and check whether the bound (1.8) remains true; a counterexample in this regime has already been reported, so a direct kernel computation would settle whether the $R_{\\mathrm{Conj}}>\\pi$ condition is necessary. Alternatively, on any $Y$ with a geodesic loop of length $d\\in(\\pi,\\pi+\\epsilon)$, evaluate $I_{GD}$ from (4.8) and verify that the cancellation in Lemma 4.4 persists with the full sum over $D(y_1,y_2)$.","tokens_in":1982,"feed_emoji":"🌊","tokens_out":2010,"duration_ms":65220,"temperature":0.7,"pith_summary":"This paper proves that the Schr\\\"odinger propagator and the half-wave propagator on an $n$-dimensional product cone satisfy the same pointwise decay in time as in Euclidean space, provided the cross-section $Y$ has no conjugate point pair within distance $\\pi$. The estimate is uniform in the radial variables and involves a factor $(r_1r_2/2t)^{-(n-2)/2+\\nu_0}$ only when $r_1r_2/(2|t|)\\ll 1$, where $\\nu_0$ is the positive square root of the smallest eigenvalue of the angular operator $P=\\Delta_h+V_0+(n-2)^2/4$. If true, this gives pointwise dispersive bounds for general closed cross-sections and identifies $\\pi$ as a natural threshold separating decay from counterexamples on spheres of radius smaller than 1.","feed_headline":"Euclidean decay holds if no conjugate pair sits within distance pi.","feed_subtitle":"On any product cone with this geometric condition, Schrodinger and wave kernels decay like flat space.","key_machinery":"The engine is a modified Hadamard parametrix for $\\cos(s\\sqrt P)$ and for the Poisson-wave operator $e^{(-s\\pm i\\pi)\\sqrt P}$ on $Y$, valid for $0\\le s\\le \\pi$ even when the exponential map is not injective. Proposition 3.2 uses $R_{\\mathrm{Conj}}>\\pi$ to make $\\exp_{y_0}$ a local covering map on a ball of radius $\\pi+\\epsilon$, and Corollary 3.6 writes the cosine kernel as a finite sum over the distance spectrum $D(y_1,y_2)$ of geodesics of length below $\\pi+\\epsilon$. Lemma 3.7 matches the jets of the two parametrices at $s=\\pi$ and $\\tilde s=0$, so that the boundary singularity of the oscillatory integral $I_{GD}$ cancels; this matching identity (3.37) is what converts the exact Bessel representation of the kernel into a bounded oscillatory integral.","core_discovery":"The central claim is Theorem 1.1: under $R_{\\mathrm{Conj}}>\\pi$ and strict positivity of $P$, the Schr\\\"odinger kernel satisfies $|e^{itH}(z_1,z_2)|\\le C|t|^{-n/2}$ times $1$ when $r_1r_2/(2|t|)\\gtrsim 1$ and times $(r_1r_2/2t)^{-(n-2)/2+\\nu_0}$ when $r_1r_2/(2|t|)\\lesssim 1$. The same geometric assumption yields Besov-space decay for the half-wave propagator $e^{it\\sqrt H}$ at rate $|t|^{-(n-1)/2}$. The author's view is that the obstacle to such estimates is precisely the presence of conjugate points within travel time $\\pi$ of the rescaled geodesic flow on the cone, so the radius-$\\pi$ threshold is the natural dividing line.","pith_inferences":["The paper explicitly leaves $R_{\\mathrm{Conj}}\\le\\pi$ to future work; the natural conjecture is failure of the global pointwise bound on cones over spheres of radius $\\le 1$, consistent with the cited counterexample in the literature.","The matching condition (3.37) suggests that microlocalized decay may survive even at the threshold, so the sharp transition might be visible only in the global kernel, not in frequency-localized pieces.","The radial factor in (1.8) is determined by the smallest angular eigenvalue $\\nu_0$; one could test the same identity for $Y=S^{n-1}$ with an inverse-square potential and compare with known Euclidean inverse-square results.","Since the proof treats the top-order singularity through the distance spectrum, a concrete computation of $I_{GD}$ on a flat torus, where $R_{\\mathrm{Conj}}=\\infty$ but $inj(Y)$ is finite, would isolate the role of geodesic loops versus conjugate points."],"forward_implications":["For any closed $Y$ with $R_{\\mathrm{Conj}}>\\pi$, the free Schr\\\"odinger decay $\\|e^{itH}\\|_{L^1\\to L^\\infty}\\le C|t|^{-n/2}$ holds when the angular operator is positive.","If the angular ground state satisfies $\\alpha\\ge 0$, one gains extra radial weight decay, $\\|r_1^{-\\alpha}e^{itH}r_2^{-\\alpha}\\|_{L^1\\to L^\\infty}\\le C|t|^{-n/2-\\alpha}$.","For negative $\\alpha$, the propagator still obeys $L^{q'}\\to L^q$ decay for $q<q(\\alpha)$, with the range restricted by the angular ground state.","The half-wave propagator satisfies $\\|e^{it\\sqrt H}f\\|_{L^\\infty}\\le C|t|^{-(n-1)/2}\\|f\\|_{\\dot B^{(n+1)/2}_{1,1}}$, the natural conical analogue of Euclidean wave decay.","Because the threshold is $\\pi$, the same method is expected to break down exactly when $R_{\\mathrm{Conj}}\\le\\pi$."],"supporting_citations":[{"why":"Proves the Schr\\\"odinger dispersive estimate on flat Euclidean cones, the model case that this paper generalizes.","marker":"[12]"},{"why":"Earlier pointwise dispersive estimates on product cones with angular-momentum dependence, which this paper improves to a uniform estimate under $R_{\\mathrm{Conj}}>\\pi$.","marker":"[26]"},{"why":"Shows the Schr\\\"odinger propagator can fail the classical pointwise estimate at conjugate points, motivating the threshold assumption.","marker":"[17]"},{"why":"Constructs resolvent and spectral measure kernels for flat cones and verifies the Blair-Ford-Marzuola conjecture, providing the flat-cone model for the present parametrix.","marker":"[54]"},{"why":"Reports failure of dispersive estimates on product cones over spheres of radius $<1$, giving the counterexample side of the $\\pi$ threshold.","marker":"[48]"},{"why":"Supplies the heat kernel estimate used for Littlewood-Paley theory in the later sections.","marker":"[19]"},{"why":"Conjectured and partially proved wave decay on flat cones, which Theorem 1.11 extends.","marker":"[4]"}],"fun_headline_variants":["Cones with conjugate radius > π attain flat-space decay rates","Schrodinger and wave decay on cones: conjugate radius > π suffices","Pointwise decay on product cones if base conjugate radius exceeds π","Euclidean decay rates on cones when conjugate radius > π","Threshold π: cone propagators decay like flat space"],"cache_read_input_tokens":51712,"weakest_assumption_plain":"The whole argument stands on the modified Hadamard parametrix staying valid up to time $s=\\pi$ when the exponential map of $Y$ is only a local covering map, not a diffeomorphism; in particular, on every geodesic sheet in the distance spectrum $D(y_1,y_2)$ the same parametrix and jet-matching must hold, while the paper writes out the details only for the sheet $d=d_h(y_1,y_2)$ and asserts the other sheets are 'the same, in fact simpler'.","fun_headline_variants_meta":{"raw":{"variants":["Cones with conjugate radius > π attain flat-space decay rates","Schrodinger and wave decay on cones: conjugate radius > π suffices","Pointwise decay on product cones if base conjugate radius exceeds π","Euclidean decay rates on cones when conjugate radius > π","Threshold π: cone propagators decay like flat space"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001064,"raw_usage":{"total_tokens":4456,"prompt_tokens":936,"completion_tokens":3520,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":3433}},"tokens_in":552,"tokens_out":3520,"duration_ms":24740,"temperature":1.0,"reasoning_tokens":3433,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:38:17.586817+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Schr\\\"odinger kernel (2.1) on the cone over $Y=S^{n-1}_\\sigma$ with $\\sigma<1$ under the same positivity assumption on $P$, and check whether the bound (1.8) remains true; a counterexample in this regime has already been reported, so a direct kernel computation would settle whether the $R_{\\mathrm{Conj}}>\\pi$ condition is necessary. Alternatively, on any $Y$ with a geodesic loop of length $d\\in(\\pi,\\pi+\\epsilon)$, evaluate $I_{GD}$ from (4.8) and verify that the cancellation in Lemma 4.4 persists with the full sum over $D(y_1,y_2)$.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the Schr\\\"odinger dispersive estimate on flat Euclidean cones, the model case that this paper generalizes."},{"cited_title":"Keeler and J","cited_arxiv_id":null,"evidence_quote":"Earlier pointwise dispersive estimates on product cones with angular-momentum dependence, which this paper improves to a uniform estimate under $R_{\\mathrm{Conj}}>\\pi$."},{"cited_title":"Hassell and J","cited_arxiv_id":null,"evidence_quote":"Shows the Schr\\\"odinger propagator can fail the classical pointwise estimate at conjugate points, motivating the threshold assumption."},{"cited_title":"Zhang , Resolvent and spectral measure for Schrodinger operators o n flat Euclidean cones , J","cited_arxiv_id":null,"evidence_quote":"Constructs resolvent and spectral measure kernels for flat cones and verifies the Blair-Ford-Marzuola conjecture, providing the flat-cone model for the present parametrix."},{"cited_title":"Huang and J","cited_arxiv_id":null,"evidence_quote":"Supplies the heat kernel estimate used for Littlewood-Paley theory in the later sections."}],"review_version":1}