{"id":"15ddb760-ce08-4c20-85a6-d0fff5086295","arxiv_id":"1908.08255","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An explicit inversion formula is derived for the weighted conical Radon transform over cones with horizontal axes and vertices on a straight line.","lead":"The paper derives an explicit formula that recovers a 3D function from its integrals over cones whose vertices lie on a line and whose axes are horizontal. The result gives an exact reconstruction method for Compton camera imaging with line detectors.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.5 drops boundary terms when converting the Gindikin inversion integral to ψ∈[0,π]; the omitted terms carry horizontal-ray data not present in T_k.","rationale":"The reader correctly noted that the principal-value interpretation and the extension of T_k to ψ∈[0,π] are left implicit. Those are genuine technical gaps and support a CONDITIONAL verdict. However, the stress-test pass found a more serious issue in the derivation of Lemma 2.5. The vertical-slice inversion formula of Theorem 2.3 is distributional: Γ is continuous on [-1,1] and jumps to 0 outside, so ∂_q Γ has endpoint delta terms. When the proof passes from the q-integral to the ψ-integral, it keeps only the classical derivative inside the interval and omits these deltas. The omitted terms involve χ_k f along the horizontal directions ±e_β, which are not encoded in the conical data T_k(ψ) for ψ∈(0,π/2) as ordinary functions; they appear only through the endpoint behavior of Γ. Because these terms are generically nonzero and do not obviously cancel after the γ, β, and p integrations, the inversion formula as displayed is not established. A direct numerical test of Lemma 2.5 with and without the endpoint correction would settle the matter. Since the central formula depends on this step, the appropriate verdict is REJECT rather than CONDITIONAL: the proof contains an identified invalid transformation, and the stated theorem is likely false as written. This is not an attack on the authors' intent; if the endpoint terms do cancel or can be absorbed into the principal-value formalism with an additional multiplicative factor, the numerical test would reveal a more benign reading, and the verdict could be revisited.","tokens_in":5969,"tokens_out":45171,"duration_ms":462161,"concrete_test":"Take k=1 and a nonnegative compactly supported Gaussian f centered at (1,0,0). For a fixed point (z,φ,η), compute the left side of Lemma 2.5, (χ_0 f)_e(z,φ,η), by direct numerical integration, and compute the right side of formula (4) with a high-accuracy quadrature that treats the ψ and p integrals as principal values. Then compute the formula with the explicit endpoint correction |cos η|/(4π) ∫_0^{2π} [ χ_1 f(z,-e_β)/(1 + sin η cos(β−φ)) + χ_1 f(z,e_β)/(1 - sin η cos(β−φ)) ] dβ added. If the uncorrected right side differs from the left side by approximately this correction while the corrected version matches, the boundary terms are real and Theorem 2.6 is false as stated.","verdict_should_be":"REJECT","load_bearing_attack":"In the proof of Lemma 2.5, the Gindikin inversion formula is applied to Γ(χ_k f)_e, which is extended by 0 for |t|>1 and takes the values (χ_k f)_e(z, ±e_β) at t=±1. The distributional derivative ∂_q Γ therefore contains delta contributions at the endpoints q = ±1 - t_0, where t_0 = sin η cos(β−φ). The line 'Since Γ(χ_k f)_e(.,t)=0 when |t|>1' is used to restrict to t∈[-1,1] and change variables to cos ψ = t_0 + q, ψ∈[0,π]. This change only retains the classical interior derivative ∂_ψ Γ and silently discards the endpoint deltas. The discarded contribution for each β is χ_k f(z, -e_β)/(-1 - t_0) - χ_k f(z, e_β)/(1 - t_0), i.e. it involves weighted X-ray data along the horizontal directions ±e_β. These horizontal-ray integrals are not recoverable from T_k restricted to ψ∈(0,π/2) as a direct limit of the displayed integrand. For a generic compactly supported f, e.g. a positive bump centered on the x-axis, χ_1 f(0, e_x) is nonzero, so the boundary term does not vanish. Thus formula (4), and hence Theorem 2.6, is missing an additive contribution unless an unstated cancellation occurs. This is a stronger defect than the two technical points raised by the reader: it is not a matter of principal-value convention or domain extension but an omitted term in the asserted identity.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers the weighted conical Radon transform T_k(f)(z,beta,psi) over all one-sided circular cones whose vertices lie on the z-axis and whose symmetry axes are horizontal, with weight ||x-b(z)||^{k-1}. It claims an explicit inversion formula: for compactly supported smooth f in R^3, f(x) is recovered by a triple integral involving T_k and its derivatives. The strategy is to relate T_k to the vertical slice transform of the weighted X-ray transform (Lemma 2.4), to invert the vertical slice transform using Gindikin's theorem (Lemma 2.5), and then to invert the resulting V-line transform via the two-dimensional Radon transform (Theorem 2.6).","tokens_in":6309,"tokens_out":10024,"duration_ms":101216,"significance":"If the main theorem were correct, it would provide an exact inversion formula for a three-dimensional family of cones, corresponding to Compton camera data with detectors on a line, and would complement existing results for other cone manifolds. The paper is concise and the overall reduction to established inversions is appealing. The derivation has no fitted parameters, and the reliance on known inversion theorems (Gindikin's vertical slice inversion and the Radon inversion) is transparent. However, the central derivation contains a serious omitted-boundary-term issue in Lemma 2.5, and the statement of Theorem 2.6 also leaves the domain of psi and the interpretation of singular integrals unspecified. These issues are load-bearing for the claimed formula.","major_comments":[{"comment":"The passage from the q-integral to the psi-integral silently discards boundary terms. The function Gamma(chi_k f)_e is extended by zero for |t|>1, so its distributional derivative in q contains delta contributions at q = 1 - sin(eta) cos(beta-phi) and q = -1 - sin(eta) cos(beta-phi), corresponding to t = 1 and t = -1. The change of variables cos(psi) = sin(eta) cos(beta-phi) + q with psi in [0,pi] captures only the classical interior derivative and omits these endpoint deltas. The omitted contribution is of the form involving chi_k f(z, e_beta) and chi_k f(z, -e_beta), i.e. weighted X-ray data along horizontal rays. Such data are not present as an additive term in the displayed integrand of (4), and for a generic compactly supported f (for example, a positive bump on the x-axis) the omitted term is nonzero. Unless a cancellation is proved, or the boundary term is explicitly expressed through T_k (for instance via the limit of T_k/sin(psi) as psi -> 0+), formula (4) and hence Theorem 2.6 are incomplete.","section":"§2, Lemma 2.5, proof of (4)"},{"comment":"The inversion formula integrates psi over [0,pi], whereas T_k is defined only for psi in (0,pi/2). The manuscript implicitly relies on an extension of T_k to obtuse cones, presumably through the symmetry T_k(z, beta, pi-psi) = T_k(z, beta+pi, psi), which follows from the relation e_{beta+pi} = -e_beta and the definition of the cone. This extension is never stated or proved. Since the formula uses values of T_k at psi > pi/2, the statement of Theorem 2.6 is currently not well-defined and needs an explicit definition or a symmetry argument.","section":"§2, Theorem 2.6"},{"comment":"The singular integrals in the formula are not given a precise interpretation. The inner psi-integral has the kernel 1/(cos(psi) - sin(gamma) cos(beta-phi)), which vanishes for generic gamma, beta, phi, psi, and the outer p-integral has the kernel 1/(r cos(eta) + x_3 sin(eta) - p), which is a Hilbert-type singular integral after the derivatives act on T_k. The paper neither states that these are principal-value integrals nor proves that the displayed iterated integral converges in the ordinary or principal-value sense for f in C_0^infinity(R^3). This is essential for the formula to be meaningful as an inversion formula.","section":"§2, Theorem 2.6 and Lemma 2.5"}],"minor_comments":[{"comment":"The statement of Gindikin's inversion formula mixes notation: the integrand contains 1/t and derivatives with respect to t, while the proof uses 1/q and derivatives with respect to q. The variables should be made consistent.","section":"§2, Theorem 2.3"},{"comment":"The domain of T_k is given as R x [0,2pi) x (0,pi/2), but Lemma 2.4 states an identity for psi in (0,pi). Please reconcile the domain of definition with the range of psi used throughout the paper.","section":"§1, Definition of T_k"},{"comment":"The notation phi in [0,2pi] should presumably be phi in [0,2pi), and the interval notation in the statement should be made uniform.","section":"§2, Lemma 2.5"},{"comment":"The sentence 'The inversion such transform can be reduced to that of the X-ray transform' contains a grammatical error and should read 'The inversion of such a transform...'.","section":"§2, proof of Theorem 2.6"}],"recommendation":"major_revision","confidential_remarks":"The boundary-term issue in Lemma 2.5 is serious and should be checked carefully by an expert in Gindikin's spherical tomography formula. If the omitted terms cannot be recovered from T_k, the main theorem is false as stated. The manuscript also needs a careful treatment of the domain of psi and of principal values before it can be considered for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper has a novel setup—an inversion formula for the conical Radon transform over cones with horizontal axes and vertices on a line—but there is a gap in the proof of Lemma 2.5 that looks serious. I'd send it to a referee, but the referee should focus on that lemma.\n\nWhat the paper does well: the problem is well-motivated from Compton cameras with line detectors. The even/odd decomposition of the weighted X-ray transform is clean, and Lemma 2.4 is correct. The idea to compose Gindikin's vertical slice inversion with V-line inversion is a reasonable strategy, and the final formula is explicit.\n\nThe soft spot is in Lemma 2.5. When applying Theorem 2.3, the function Γ(χ_k f)_e is extended by zero for |t|>1, but it has nonzero values at t=±1. The distributional derivative therefore contains delta functions at those endpoints. In the change of variables to ψ∈[0,π], they keep only the interior derivative and discard the deltas. The omitted terms involve χ_k f(z, ±e_β), i.e., horizontal-ray integrals, which are not recoverable from T_k on ψ∈(0,π/2). For a generic bump, these terms are nonzero. So formula (4) appears to be missing an additive boundary contribution. This is not the same as the principal-value or domain-extension issues the reader flagged; those are matters of convention. This is an omitted term in an identity.\n\nMaybe the boundary terms cancel after the β and γ integrations, but no argument is given. As written, the proof of the main theorem is incomplete.\n\nThat said, the paper is not a wreck. The problem is important, the strategy is sensible, and with a careful treatment of the boundary terms—either showing they cancel or subtracting them—the result might be salvageable. The citation pattern looks fine, and the claimed novelty is real.\n\nFor peer review: I would accept it because the question is worthwhile and the gap is specific enough to be fixable. But the referee needs to check Lemma 2.5 carefully. I would not cite it in its current form.","headline":"A useful new inversion formula for a practical cone transform, but the proof of Lemma 2.5 drops boundary terms that appear to break the identity.","tokens_in":6794,"tokens_out":11380,"would_cite":false,"duration_ms":107209,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["44A12","53C65"],"pacs":[],"model":"deepseek-v4-flash","headline":"A function in 3D can be reconstructed exactly from integrals over all cones with horizontal axes and vertices on a line, for every integer weight.","keywords":["conical Radon transform","inversion formula","Compton camera imaging","vertical slice transform","V-line transform","weighted X-ray transform","spherical integral geometry","cone-beam reconstruction"],"falsifier":"Take a compactly supported smooth $f$ whose conical transform $T_k$ can be computed to high accuracy, evaluate the right-hand side of Theorem 2.6 at a point outside the support of $f$, and check that the principal-value integral is zero; a nonzero value, or divergence as the regularization is removed, would falsify the claimed inversion.","tokens_in":5780,"feed_emoji":"📐","tokens_out":15363,"duration_ms":122817,"temperature":0.7,"pith_summary":"The paper establishes an exact, explicit inversion formula for a weighted conical Radon transform in $\\mathbb{R}^3$, defined by integrating a compactly supported smooth function over all one-sided circular cones whose vertices lie on a straight line and whose axes are horizontal. The formula recovers the original function from the transform values and their derivatives by a finite combination of integrals, for every integer weight parameter $k$. This matters because this three-parameter family of cones models Compton-camera detection with detectors arranged along a line: each detected photon gives a cone of possible source directions, and an explicit inversion makes direct analytic reconstruction possible rather than iterative approximation.","feed_headline":"Cone-scan data on a line can be exactly inverted","feed_subtitle":"Triple-integral formula recovers a function from Compton-camera conical projections, for any integer weight.","key_machinery":"The proof is carried by three transforms and one connecting identity. The vertical slice transform $\\Gamma g(\\phi,t)$ integrates a function on the sphere over the circle where $e_\\phi\\cdot\\omega=t$, and its spherical inversion (Theorem 2.3) is the first reconstruction step. The weighted X-ray transform $\\chi_k f(z,\\omega)$ integrates $f$ along the ray from the vertex $b(z)$ in direction $\\omega$ with weight $r^k$. Lemma 2.4 connects them: $\\Gamma(\\chi_k f)(z,\\beta,\\cos\\psi)=T_k(f)(z,\\beta,\\psi)/(2\\pi\\sin^2\\psi)$, so cone data give slice data of the weighted X-ray transform. Because slice circles are symmetric about the horizontal plane, only the even part $(\\chi_k f)_e$ survives, and an identity expresses the unweighted even X-ray data $(\\chi_0 f)_e$ as an integral of $z$-derivatives of $(\\chi_k f)_e$, reducing any weight $k$ to the unweighted case. The even X-ray data are then read as V-line integrals in each vertical half-plane, and two-dimensional Radon inversion finishes the reconstruction.","core_discovery":"The central claim, Theorem 2.6, is that for every integer $k\\ge 1$ and every compactly supported smooth $f$, the value $f(x)$ at $x=(r\\,e_\\varphi,x_3)$ is reproduced by the displayed integral of the conical data $T_k(f)(p/\\sin\\eta,\\beta,\\psi)$ and its derivatives, with the singular kernels $1/(\\cos\\psi-\\sin\\gamma\\cos(\\beta-\\varphi))$ and $1/(r\\cos\\eta+x_3\\sin\\eta-p)$. The integrations run over the opening angle, the detector direction, and the cone-vertex position. Thus the three-dimensional family of conical projections with line vertices and horizontal axes determines $f$ exactly, not merely approximately. The proof converts cone integrals into vertical-slice data on the sphere, recovers the even weighted X-ray transform, and inverts the resulting V-line data by the two-dimensional Radon transform.","pith_inferences":["The $\\psi$-integral in Theorem 2.6 runs to $\\pi$ although $T_k$ is defined only for $\\psi\\in(0,\\pi/2)$; the paper leaves implicit that obtuse-angle data must come from the symmetry $T_k(z,\\beta,\\pi-\\psi)=T_k(z,\\beta+\\pi,\\psi)$, so a detector recording only acute cones would need additional measurements or an extrapolation step.","The two kernel denominators are singular inside the integration domain, and the paper states no principal-value prescription; any numerical implementation must choose and validate a regularization before the formula can handle measured data.","Because the proof factors through the vertical slice transform, the same strategy could plausibly invert other three-parameter cone manifolds with horizontal axes, such as vertices lying on a closed curve; the paper does not pursue those geometries.","The inversion differentiates the data $k+1$ times in $p$ and once in $\\psi$, so noise sensitivity is a likely practical obstacle; simulated-phantom tests of the formula's stability would be a natural next step."],"forward_implications":["For every positive integer $k$, a compactly supported smooth function in $\\mathbb{R}^3$ is determined exactly by its horizontal conical Radon transform on the three-parameter family of cones with vertices on a line.","Compton-camera setups with detectors on a line acquire a direct analytic reconstruction route: recover even weighted X-ray data, then invert a planar Radon transform slice by slice.","The inversion separates over the azimuth $\\varphi$, so the reconstruction can be carried out independently for each vertical half-plane.","The formula handles every integer weight $k$ with the same structure, so stronger radial weighting needs no new inversion theory."],"supporting_citations":[{"why":"Supplies the spherical inversion formula for the vertical slice transform that starts the reconstruction chain.","marker":"[9]"},{"why":"Shows the reduction of V-line inversion to X-ray inversion that the proof of Theorem 2.6 follows on each half-plane.","marker":"[3]"},{"why":"Provides the two-dimensional Radon inversion used to reconstruct the function on each vertical half-plane.","marker":"[13]"},{"why":"Supplies the identity that expresses unweighted even X-ray data from weighted X-ray data by z-derivatives.","marker":"[17]"}],"fun_headline_variants":["Exact inversion for horizontal-cone line data","Line-vertex cone projections: exact recovery theorem","Conical Radon data on a line: exact formula","Explicit exact inversion for line-cone Radon"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The derivation assumes that the singular integrals in the inversion formula are interpreted as principal values and that data for obtuse cone opening angles are known through the symmetry relation; if either assumption fails, the equality in Theorem 2.6 need not hold.","fun_headline_variants_meta":{"raw":{"variants":["Exact inversion for horizontal-cone line data","Line-vertex cone projections: exact recovery theorem","Conical Radon data on a line: exact formula","Explicit exact inversion for line-cone Radon"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000306,"raw_usage":{"total_tokens":1659,"prompt_tokens":758,"completion_tokens":901,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":374,"completion_tokens_details":{"reasoning_tokens":839}},"tokens_in":374,"tokens_out":901,"duration_ms":9094,"temperature":1.0,"reasoning_tokens":839,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:46:05.423765+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a compactly supported smooth $f$ whose conical transform $T_k$ can be computed to high accuracy, evaluate the right-hand side of Theorem 2.6 at a point outside the support of $f$, and check that the principal-value integral is zero; a nonzero value, or divergence as the regularization is removed, would falsify the claimed inversion.","supporting_citations":[{"cited_title":"Spherical tomography and spherical integral geometry","cited_arxiv_id":null,"evidence_quote":"Supplies the spherical inversion formula for the vertical slice transform that starts the reconstruction chain."},{"cited_title":"Analytical reconstruction formula for one- dimensional compton camera","cited_arxiv_id":null,"evidence_quote":"Shows the reduction of V-line inversion to X-ray inversion that the proof of Theorem 2.6 follows on each half-plane."},{"cited_title":"The radon transform, volume 2","cited_arxiv_id":null,"evidence_quote":"Provides the two-dimensional Radon inversion used to reconstruct the function on each vertical half-plane."},{"cited_title":"Analytic inversion of a conical radon transform arising in application of compton cameras on the cylinder","cited_arxiv_id":null,"evidence_quote":"Supplies the identity that expresses unweighted even X-ray data from weighted X-ray data by z-derivatives."}],"review_version":1}