{"id":"4c84bff6-5521-4261-bbf9-b1529216f1bf","arxiv_id":"1908.02554","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For potentials that are, far from the origin, functions of distance to a conical surface, the Schrödinger operator has infinitely many eigenvalues below its essential spectrum and their counting function behaves like a geometric constant times |log E|.","lead":"This paper proves that a quantum particle attracted to an infinite cone-shaped surface by a wide class of potentials has infinitely many bound states accumulating at a threshold energy. It computes the accumulation rate and shows it depends only on the cone's cross-section, not on the detailed potential.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 4 upper bound uses partition cells Omega_p that are not contained in P_{R,delta}; the printed Neumann bracketing is therefore unjustified.","rationale":"I read the paper in good faith. The central claim, a generalization of the conical-layer and delta-potential results of [OBP], is plausible and the main structure of the proof is sound. The one-dimensional preparatory propositions (2.2 through 2.8) are carefully argued; the Agmon-type estimate and the exponential eigenvalue convergence appear correct. The lower-bound proof in Section 3 has a sign typo in the final displayed constant, but the preceding computation with Proposition 2.1 gives the corrected expression, so this is not a deep issue. The most significant defect is in the upper-bound proof of Section 4: the partition Omega_p is not contained in P_{R,delta}, and the residual region Omega_{m+1} can include points near t=0 where v(t) is not bounded below by any epsilon_1 > epsilon_0. These flaws invalidate the Neumann bracketing step as written, so the paper does not currently establish the limsup bound needed for the asymptotic equality. However, both issues are localized and evidently repairable (e.g., partition in the variables (r, u=t/r) and a standard convolution argument for the tail), and the claimed constant k_S is consistent with the established results in [OBP]. The reader's weakest-assumption (isolated ground state of Q) is actually well justified by hypothesis (ii) and the one-dimensional Sturm-Liouville theory; it is not the soft spot. I agree with the reader's conditional verdict, for the same reason they cite (the Section 4 partition), rather than their stated weakest assumption.","tokens_in":17103,"tokens_out":36054,"duration_ms":342909,"concrete_test":"Verify the inclusion Omega_p subset P_{R,delta} for the printed definitions. Fix p<m, let r = r_p + (r_{p+1}-r_p)/2 and t = (t_p+t_{p+1})/2 = delta(r_p+r_{p+1})/2. Then delta r = delta r_p + delta(r_{p+1}-r_p)/2, and |t| = delta(r_p+r_{p+1})/2 = delta r + delta(r_{p+1}-r_p)/2 > delta r, so the point is outside P_{R,delta}. This confirms that Omega_p is not contained in the domain of G_{R,delta}^{[n]}. Next, check whether the quadratic-form inequality following the definition of h_{p,delta}^{[n]} is valid for all test functions in H^1(Omega_p) supported in Omega_p; if those functions are then extended to P_{R,delta} without restriction, the bracketing argument fails because the extension is not in the form domain of the original operator.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The upper-bound proof in Section 4 decomposes P_{R,delta} = {(r,t): r>R, |t|<delta r} into cells Omega_p = {r in (r_p,r_{p+1}), t in (t_p,t_{p+1})}, with r_p = R + pL/m and t_p = delta r_p. For p<m, the point (r,t) with r = r_p + epsilon and t = delta r_{p+1} - epsilon lies in Omega_p but has |t| > delta r, because delta r_{p+1} > delta r when r = r_p. Hence Omega_p is not a subset of P_{R,delta}, and the inequality g_{R,delta}^{[n]} >= direct sum h_{p,delta}^{[n]} used for Neumann bracketing is not justified. Moreover, the residual region Omega_{m+1} contains points with arbitrarily small |t|, including t = 0 for r > R+L, where v(t) may lie below the chosen level epsilon_1, so the claim N_{epsilon_0-E}(h_{m+1,delta}^{[n]}) = 0 is not supported by assumption (iii). A correct proof would need a different partition, for instance using the variable u = t/r so that the cells respect the cone boundary. As printed, the limsup bound N_{epsilon_0-E}(H) <= k_S|log E| is not established. Separately, the final display of the lower bound in Section 3 has a sign error: the correct expression from Proposition 2.1 is sqrt((-lambda_m - C(delta+epsilon)/4)_+), not sqrt((lambda_m - C(delta+epsilon)/4)_+); this is an evident typo, not a structural gap.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the three-dimensional Schrödinger operator H = -Δ + V on R^3 with V(x) = v(d_S(x)) + w(x), where S is an infinite conical surface with C^4-smooth cross-section Σ (not a plane), v is an even one-dimensional potential with an isolated ground state ε_0 below v_∞ := liminf v, and w = o(|x|^{-2}). The main result, Theorem 1.1, states that the essential spectrum of H is [ε_0, ∞), that the discrete spectrum is infinite below ε_0, and that the eigenvalue counting function obeys N_{ε_0-E}(H) ≃ k_S |log E| as E → 0+, with k_S = (1/2π) Σ_{λ_j(K_S)<0} √(-λ_j(K_S)), where K_S = -d^2/ds^2 - κ^2/4 on the cross-section loop. The proof follows the strategy of Ourmières-Bonafos and Pankrashkin [OBP]: it establishes one-dimensional Agmon-type and eigenvalue-convergence estimates for truncated operators, then uses tubular-coordinate changes and min-max bracketing to reduce the counting problem to the one-dimensional Kirsch-Simon result. A Weyl-sequence argument identifies the essential spectrum.","tokens_in":17387,"tokens_out":7673,"duration_ms":76400,"significance":"If the proof is completed, the result is significant: it shows that the |log E| asymptotics discovered for Dirichlet conical layers and δ-interactions on conical surfaces are not artifacts of those particular models but hold for a large class of attractive potentials depending on distance to the cone, with the same geometric constant k_S. The main theorem is parameter-free: k_S is computed solely from the geodesic curvature of the cross-section, and the perturbation w is allowed to be any o(|x|^{-2}) term. The paper also contains useful technical ingredients, such as uniform Agmon estimates and exponential eigenvalue convergence for truncated one-dimensional operators, which are of independent interest. The reliance on [OBP] is substantial, but the present result genuinely extends the framework to general v. The proofs are largely standard and the central claim is very plausible.","major_comments":[{"comment":"The partition used for the Neumann bracketing in the upper bound is not contained in P_{R,δ}. The cells are defined as Ω_p = {(r,t): r ∈ (r_p, r_{p+1}), t ∈ (t_p, t_{p+1})} with t_p = δ r_p. For r = r_p + ε and t = δ r_{p+1} - ε' with ε, ε' > 0 sufficiently small, the point (r,t) lies in Ω_p but satisfies |t| > δ r, because δ r_{p+1} - δ r_p = δ L/m > ε'(1+δ) for ε' < δ L/(m(1+δ)). Hence Ω_p ⊄ P_{R,δ}, and the asserted inequality g_{R,δ}^{[n]} ≥ ⊕_{p=0}^{m+1} h_{p,δ}^{[n]} is not justified, since the forms h_{p,δ}^{[n]} are defined on domains extending outside the original domain. In addition, the residual region Ω_{m+1} contains, for any R, the strip {r > R, |t| < δ R}, including points with t = 0; the claim that v(t) ≥ ε_1 > ε_0 on Ω_{m+1} is therefore unsupported by assumption (iii), and the conclusion N_{ε_0-E}(h_{m+1,δ}^{[n]}) = 0 does not follow. The limsup bound N_{ε_0-E}(H) ≤ k_S |log E| is consequently not established as printed. A corrected argument should partition P_{R,δ} with cells that respect the cone boundary |t| = δ r, for example by using the variable u = t/r.","section":"Section 4, definition of Ω_p and Ω_{m+1}"}],"minor_comments":[{"comment":"The final display of the lower bound contains a sign error: the factor under the square root should be (-λ_m(K_S) - C(δ+ε)/4)_+, not (λ_m(K_S) - C(δ+ε)/4)_+. As printed, the expression vanishes for the relevant negative eigenvalues λ_m(K_S), making the lower bound trivial. This is evidently a typo, because the preceding line has the correct sign in the quadratic form g_{R,δ}^{[m]}, but it must be corrected for the proof to be read coherently.","section":"Section 3, final display"},{"comment":"In the display defining D(g_{R,δ}), the notation '∂rv, ∂tv' should read '∂_r φ, ∂_t φ' or '∂_r v, ∂_t v' with a consistent symbol; as written it is a typographical error.","section":"Section 4, form domain of g_{R,δ}"},{"comment":"The Weyl-sequence computation writes 'V H φ_n' where V is the unitary change of variables; the expression should be 'V H V^{-1} (V φ_n)' or 'V H φ_n' with φ_n already in the transformed coordinates. The intended meaning is clear, but the notation is imprecise.","section":"Section 5, Eq. (5.1)"},{"comment":"Several computations are said to follow 'almost literally' from [OBP]. This is acceptable, but the paper would be more self-contained if the few modifications (e.g., the presence of v(t) and the ε/|x|^2 corrections) were spelled out in one displayed equation each, rather than referring to the companion paper for the bulk of the estimates.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a serious contribution and the central claim is very likely correct. The main obstacle is the Section 4 upper-bound partition, which as printed invalidates the Neumann bracketing; however, this is a fixable technical issue (e.g., using the variable u = t/r), so I recommend major revision rather than rejection. The paper shares an author with [OBP], which is used as a technical template; this is not improper, but the editor may wish to ensure the overlap is transparent. The sign error in Section 3 is minor and should be corrected in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one over the weekend. The result is a genuine extension: one counting law, k_S|log E|, for any even v with an isolated ground state below v_infinity, covering the hard-wall layer and delta-interaction as special cases. The one-dimensional machinery in Section 2 is solid and will be useful beyond this paper. I think the theorem is true.\n\nTwo problems in the written proof.\n\nSection 3: the final display has a sign error in front of lambda_m. Applying Proposition 2.1 gives sqrt((-lambda_m - C(delta+epsilon)/4)_+), not sqrt((lambda_m - C(delta+epsilon)/4)_+). As printed, the argument is negative for the modes you need, so the lower bound would vanish. Evidently a typo.\n\nSection 4: the upper-bound partition is not a partition of P_{R,delta}. The cells Omega_p = {r in (r_p,r_{p+1}), t in (delta r_p, delta r_{p+1})} leave the cone: for r just above r_p, t up to delta r_{p+1} exceeds delta r. Omega_m with r > r_m is worse. So g_{R,delta} >= direct sum h_{p,delta} is unjustified. And the residual region Omega_{m+1} includes r > R+L with |t| small, down to t=0, where v(t) may lie below the epsilon_1 chosen from assumption (iii); the assertion that h_{m+1} contributes zero is not supported. This is a real gap, not a typo, though I expect a corrected partition, for instance in the variable u=t/r, would restore the [OBP] argument.\n\nEverything else checks out: no fitted constants, k_S is geometric, the citation to [OBP] is appropriate, and the isolated-ground-state condition is natural.\n\nBottom line: worth serious referee time. I would want to see the corrected Section 4 before citing it myself.","headline":"A genuine and probably true generalization of the conical bound-state counting law, but the printed upper-bound proof has a real partition gap that needs fixing; the lower-bound typo is easy.","tokens_in":17986,"tokens_out":16429,"would_cite":true,"duration_ms":170991,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P20","81Q10"],"pacs":[],"model":"deepseek-v4-flash","headline":"A potential depending only on distance to a non-planar conical surface forces infinitely many eigenvalues below the essential spectrum, with a universal logarithmic accumulation rate.","keywords":["conical surface","Schrödinger operator","infinitely many bound states","essential spectrum","eigenvalue counting function","geodesic curvature","logarithmic asymptotics"],"falsifier":"Pick a circular cone and an admissible well $v$ with an isolated bound state, compute the first several thousand eigenvalues of $H$ below $\\varepsilon_0$ in tubular coordinates, and check whether $N_{\\varepsilon_0-E}(H)/(k_S|\\log E|)$ tends to $1$ as $E\\to0^+$; a different finite limit, or only finitely many discrete eigenvalues, would disprove the theorem.","tokens_in":16835,"feed_emoji":"⚛️","tokens_out":10038,"duration_ms":92140,"temperature":0.7,"pith_summary":"The paper proves a universality statement about bound states near attractive conical surfaces in three dimensions. It considers Schrödinger operators $H=-\\Delta+V$ on $\\mathbb{R}^3$ whose potential, up to a rapidly decaying remainder, is a one-dimensional function $v$ of the distance to an infinite cone $S$. If the one-dimensional operator $-d^2/dx^2+v$ has an isolated ground state $\\varepsilon_0$ strictly below the limit of $v$ at infinity, then the essential spectrum of $H$ begins at $\\varepsilon_0$ and infinitely many discrete eigenvalues accumulate at $\\varepsilon_0$. The eigenvalue count near threshold satisfies $N_{\\varepsilon_0-E}(H)\\sim k_S|\\log E|$, where $k_S$ is a positive number computed from the geodesic curvature of the cone's cross-section. This makes the logarithmic accumulation previously seen for hard-wall conical layers and delta-interactions a generic phenomenon that does not depend on the detailed shape of the attractive potential.","feed_headline":"Attractive cones force infinitely many quantum bound states","feed_subtitle":"Bound states pile up at a rate set only by the cone's curvature, not by the potential's shape.","key_machinery":"The load-bearing object is the geometrically induced one-dimensional operator $K_S=-d^2/ds^2-\\kappa^2/4$ on the cross-section loop $\\Sigma=S\\cap S^2$, where $\\kappa$ is the geodesic curvature of the loop. Its negative eigenvalues define $k_S$, the constant controlling the counting asymptotics, and $k_S>0$ exactly when $\\Sigma$ is not a great circle. The proof also uses two one-dimensional tools: the operator $Q=-d^2/dx^2+v$, whose isolated ground state $\\varepsilon_0$ sets the threshold, and a classical half-line counting estimate that turns effective attractive $-1/r^2$ terms into the $|\\log E|$ law. These are combined through tubular coordinates around the cone, operator bracketing, and an exponential decay estimate for the one-dimensional ground state, which lets the full three-dimensional problem be compared with direct sums of one-dimensional operators.","core_discovery":"The central claim is Theorem 1.1: for $V(x)=v(d_S(x))+w(x)$ with $w(x)=o(|x|^{-2})$, where $v$ is even, locally integrable, bounded below, and the one-dimensional operator $Q=-d^2/dx^2+v$ has an isolated ground state $\\varepsilon_0<\\liminf_{x\\to\\infty}v(x)$, the operator $H=-\\Delta+V$ on $\\mathbb{R}^3$ has essential spectrum $[\\varepsilon_0,\\infty)$, infinitely many discrete eigenvalues, and counting asymptotics $N_{\\varepsilon_0-E}(H)\\sim k_S|\\log E|$ as $E\\to 0^+$. The constant is $k_S=(2\\pi)^{-1}\\sum_{\\lambda_j(K_S)<0}\\sqrt{-\\lambda_j(K_S)}$, with $K_S=-d^2/ds^2-\\kappa^2/4$ acting on the cross-section loop. Because the same $k_S$ appears for Dirichlet conical layers and for delta-potentials on conical surfaces, the theorem identifies the logarithmic law as a universal spectral-geometric effect rather than a feature of those specific interactions.","pith_inferences":["Since only the far-field behavior of $v$ and the cross-section geometry enter, the same mechanism should produce a similar logarithmic law for conical surfaces in higher dimensions with $(n-2)$-dimensional cross-sections; the paper does not treat that case.","The argument suggests that any transverse confinement with an isolated threshold plus an effective attractive $-1/r^2$ far field should give the same count; one testable extension would replace $v(d_S(x))$ by a magnetic or metric mechanism producing the same effective potential.","The assumption $w=o(|x|^{-2})$ is likely close to sharp, and probing slower decay could reveal whether the constant $k_S$ changes or the asymptotic law breaks.","The leading term carries no information about the shape of $v$; a natural next step is to compute the next-order correction, which would depend on the detailed potential and distinguish models with identical $k_S$."],"forward_implications":["For any admissible attractive one-dimensional profile $v$, a non-planar conical surface $S$ forces infinitely many bound states below the essential spectrum.","The bottom of the continuous spectrum is exactly the isolated ground-state energy of $Q$, independently of the cone's geometry.","The counting prefactor $k_S$ is universal: hard-wall layers, delta-interactions, and general short-range wells around the same cone all give the same $|\\log E|$ rate.","The result is stable under rapidly decaying perturbations $w=o(|x|^{-2})$, so the logarithmic law persists when the cone is only approximately attractive.","A Weyl sequence construction places every energy $\\varepsilon_0+k^2$, $k\\ge 0$, in the essential spectrum, so the entire half-line $[\\varepsilon_0,\\infty)$ is covered."],"supporting_citations":[{"why":"Supplies the earlier result for conical layers and delta-interactions whose proof is adapted, including the tubular-coordinate decomposition and the definition of $k_S$.","marker":"[OBP]"},{"why":"Provides the one-dimensional half-line eigenvalue-counting estimate used to obtain the $|\\log E|$ asymptotics.","marker":"[KS]"},{"why":"Introduces the conical-layer model whose infinite discrete spectrum and logarithmic count are here generalized to arbitrary attractive potentials.","marker":"[ET]"}],"fun_headline_variants":["Attractive cones spawn infinite bound states","Cone curvature dictates quantum bound-state pileup","Infinite bound states on cones with universal log count","Cone curvature yields infinite bound states, logarithmic rate"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result depends on the one-dimensional potential $v$ having a bound state whose energy lies strictly below the value the potential approaches at infinity; if that energy gap closes, the exponential localization and the logarithmic counting law can fail.","fun_headline_variants_meta":{"raw":{"variants":["Attractive cones spawn infinite bound states","Cone curvature dictates quantum bound-state pileup","Infinite bound states on cones with universal log count","Cone curvature yields infinite bound states, logarithmic rate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001548,"raw_usage":{"total_tokens":6173,"prompt_tokens":914,"completion_tokens":5259,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":530,"completion_tokens_details":{"reasoning_tokens":5200}},"tokens_in":530,"tokens_out":5259,"duration_ms":39886,"temperature":1.0,"reasoning_tokens":5200,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:42:06.916114+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Pick a circular cone and an admissible well $v$ with an isolated bound state, compute the first several thousand eigenvalues of $H$ below $\\varepsilon_0$ in tubular coordinates, and check whether $N_{\\varepsilon_0-E}(H)/(k_S|\\log E|)$ tends to $1$ as $E\\to0^+$; a different finite limit, or only finitely many discrete eigenvalues, would disprove the theorem.","supporting_citations":[],"review_version":1}