{"id":"27cfe297-ec8b-498a-98ea-7f68a5199a77","arxiv_id":"2412.00822","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The low-intensity Poisson-Voronoi tessellation of H2 × H2 with the L1 metric converges to an isometry-invariant ideal tessellation whose cell ends are unions of boundary circles, and equal-separation loci between corona points are unbounded almost surely.","lead":"This paper constructs the ideal Poisson-Voronoi tessellation, a random partition of space generated by a very sparse random set of points, on the product of two hyperbolic planes with the L1 metric. It proves the partition is invariant under all isometries and describes the shape of its cells at infinity.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.1's corona identification rests on an unproved assertion about mixed-boundary nuclei; if false, the entire IPVT construction changes.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing point: the exclusion of mixed-boundary nuclei in the proof of Theorem 2.1 is asserted rather than demonstrated. This is the foundation of the whole paper: if the corona measure were not supported on ∂H2 × ∂H2 × R≥0, the separation formula (Proposition 2.2), the isometry action (Corollary 3.2), and the cell-boundary description (Theorem 1.2) would all change. The concern is not that the claim is false — a direct volume computation indicates the expected number of such nuclei in any bounded-delay window is O(1/log(1/λ)) and therefore vanishes — but that the paper does not provide that computation. Since the proof is a single line referring to (2.2), the gap is real and fixable. It does not warrant rejection, because the underlying Poisson-process calculation is standard and likely correct; it does warrant a conditional acceptance with a request for the missing volume estimate. The reader already gave CONDITIONAL, so the verdict remains unchanged. I did not identify a stronger objection: the isometry-group lemma appears correct, the covering comparison with Biermé–Estrade is terse but plausible, and the algebraic calculation in Theorem 1.3 is omitted but can be independently verified. The corona identification is the most load-bearing because it is the unique point whose failure would invalidate the central construction rather than merely require additional detail.","tokens_in":10138,"tokens_out":14246,"duration_ms":127571,"concrete_test":"Compute explicitly the expected number of nuclei of X(λ) in the annulus {R_λ − C ≤ d(o,·) ≤ R_λ + C} with d(o1, ·) ≤ A, using φ(r) = 2π^2(r cosh r − sinh r) and the product structure of Vol_M. If this expectation is O(C/log(1/λ)) for every fixed A, C, then the mixed-boundary contribution vanishes in the limit and the corona identification of Theorem 2.1 is justified. If instead the bound is O(1) or worse, Theorem 2.1 must be revised. The analogous computation with d(o2, ·) ≤ A should also be checked.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction in Theorem 2.1 depends on the one-sentence assertion after (2.2) that 'any nuclei converging in the Gromov sense towards (H2 × ∂H2) or (∂H2 × H2) would have a.s. infinite delay.' This is the only argument excluding those mixed-boundary components from the corona, and it is not justified by any computation. In fact, the phrasing is misleading: a nucleus with bounded first coordinate d(o1, X_i1) ≤ A and second coordinate near R_λ = log(1/λ) − loglog(1/λ) has proto-delay D_i^(λ) = d(o,X_i) − R_λ = O(1), not infinite. Such nuclei are rare only because their expected count vanishes: the shell {R_λ ≤ d(o,·) ≤ R_λ + C} has volume ~ π^2 R_λ e^{R_λ} C, while the subset with d(o1,·) ≤ A has volume ~ Vol(B_A) π e^{R_λ} C, so after multiplying by λ the mean is Vol(B_A) π C / log(1/λ) → 0. A similar computation holds for the second coordinate. If this volume estimate failed, or if one only proved a weaker decay, the corona measure in (2.4) would require extra atoms on (H2 × ∂H2) ∪ (∂H2 × H2), changing Proposition 2.2, Corollary 3.2, and every subsequent geometric statement. The paper does not supply this verification, nor does it check the remaining hypotheses of [DCE+23, Theorem 2.3] for this non-hyperbolic space.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs an ideal Poisson–Voronoi tessellation (IPVT) for M = H2 × H2 equipped with the L1 metric. The main result (Theorem 2.1) identifies the low-intensity limit of Voronoi diagrams of a Poisson point process on M with a corona Poisson process on ∂H2 × ∂H2 × R≥0, relying on the abstract convergence theorem of [DCE+23]. The paper then proves isometry invariance of IPVT(M), describes the boundary behavior of cells in terms of ends, and studies the separation process along a geodesic to a boundary point, including a tie-break probability. The manuscript is clearly written and the overall strategy is coherent, but several load-bearing steps in the proofs are abbreviated or contain an incorrect-sounding assertion.","tokens_in":10392,"tokens_out":8619,"duration_ms":85949,"significance":"If the results are correct, this is a meaningful extension of IPVT theory beyond hyperbolic spaces: M is neither δ-hyperbolic nor a Riemannian symmetric space, yet it admits a product Poisson-kernel separation formula, a transitive corona action, and computable cell geometry. The explicit formulas for the no man's land and the separation process are useful tools and parallel recent work of Fra¸czyk–Mellick–Wilkens. The paper also benefits from building on the independently established machinery of [DCE+23]. However, the current manuscript leaves several central verifications to one-sentence arguments or to 'tedious but elementary' calculations, so the proofs are not yet at the standard required for the main theorems.","major_comments":[{"comment":"The assertion that any nuclei converging in the Gromov sense towards (H2 × ∂H2) or (∂H2 × H2) would have a.s. infinite delay is not correct as written: a nucleus with bounded first coordinate, say d(o1, x1) ≤ A, and second coordinate at distance R_λ − d(o1, x1) + O(1) from o2 has proto-delay D_i^(λ) = O(1), not infinite. The conclusion is nevertheless true, but the correct argument is a shell-volume estimate: for fixed A and a bounded delay window, the expected number of such nuclei is O(λ e^{R_λ} Vol(B_A)) = O(Vol(B_A)/log(1/λ)) → 0. This estimate is absent. Since this is the step that excludes the mixed-boundary components and thereby determines the corona measure in (2.4), it must be supplied explicitly. The proof should also verify the remaining hypotheses of [DCE+23, Theorem 2.3] for this non-hyperbolic space rather than citing it in one line.","section":"§2, proof of Theorem 2.1, after Eq. (2.2)"},{"comment":"The covering argument for the deposition model of hyperbolic crosses is compressed into the sentence that the Biermé–Estrade process covers P a.s. by 'many small balls' and hence so does the model with large disks excluded. This is load-bearing for Theorem 1.2. The conditional intensity in (4.3) indeed differs from the Biermé–Estrade process by removing disks with ρ^4 ≥ (1+x^2)(1+y^2), but the paper does not prove that this truncation does not destroy the covering property. A precise check is needed, for example by showing that excluded disks contribute only finitely many covering balls locally, or by verifying the covering criterion directly for the truncated intensity.","section":"§4, proof of Theorem 1.2, around Eq. (4.3)"},{"comment":"The derivation of the limiting intensity measure ν∞(η, ξ) is omitted entirely. After the phrase 'After some tedious but elementary calculation' the paper jumps to the formula stated in Theorem 1.3. This intensity is the main quantitative result of the second part of the paper and also feeds into Corollary 1.5, so the calculation should be provided in detail or at least in an appendix. Without it, the convergence of the rescaled separation process is not verifiable from the manuscript.","section":"§5, proof of Theorem 1.3(ii)"}],"minor_comments":[{"comment":"The sentence 'Θ = (Θ1, . . .) and Φ = (Φ1, . . .) are i.i.d. uniform on ∂H2 × ∂H2' is a typo: each Θ_i and each Φ_i should be uniform on ∂H2, not on the product ∂H2 × ∂H2.","section":"Theorem 2.1"},{"comment":"In the definition of E(Θ1, Φ1), the second union is written as {(τ1, Φ1) ; τ1 ∈ H2}; it should be τ1 ∈ ∂H2.","section":"Theorem 1.2 proof"},{"comment":"The caption uses the notation S∞(e^{-x}, e^{-x}) without defining it; please either define this process explicitly or rephrase the caption.","section":"Figure 1.1"},{"comment":"The variables θ and ϕ are used both as boundary parameters and as angular coordinates in the cosine expansion; a different notation for the angular variables would improve readability.","section":"Eq. (5.1)"},{"comment":"The final step of the proof, in which the exponential map is used to show uniqueness of the isometry from its derivative, is very brief; a sentence explaining why the exponential map is surjective in this Finsler setting would help the reader.","section":"Lemma 3.1"}],"recommendation":"major_revision","confidential_remarks":"The central construction is plausible and the missing shell-volume calculation for mixed-boundary nuclei is straightforward to add, so I would not reject the paper. The main issue is that Theorem 2.1, the cornerstone of the paper, currently rests on an incorrect one-sentence justification; this must be corrected and the hypotheses of [DCE+23, Theorem 2.3] must be checked explicitly. The other major comments concern omissions of verification for covering and intensity limits. Once those are added, the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does what it says: it builds the ideal Poisson–Voronoi tessellation on H2 × H2 with the L1 metric, proves isometry invariance, and analyzes cell boundaries and separation asymptotics. That is a new example beyond hyperbolic spaces, and the product structure of the separation formula is clean and appealing. The isometry group computation (Lemma 3.1) is a useful piece of folklore made explicit, and the hyperbolic-cross deposition model in Section 4 is a nice visual and technical tool. The paper follows the abstract recipe of [DCE+23] honestly; relying on that theorem is not circular, since it is proven elsewhere and no parameters are fitted here.\n\nThe main soft spot is in the proof of Theorem 2.1. The one-sentence assertion that nuclei converging to (H2 × ∂H2) or (∂H2 × H2) have 'a.s. infinite delay' is not justified and is worded inaccurately. A nucleus with bounded first coordinate and second coordinate near R_λ has proto-delay O(1), not infinite. What actually saves the argument is that the expected number of such nuclei vanishes, roughly like C/log(1/λ), as the stress-test note computes. The paper omits that volume estimate entirely. The fix is straightforward: add a short calculation showing the mean count of mixed-boundary nuclei tends to zero, and then the corona converges to ∂H2 × ∂H2. This is a rigor gap, not a fatal flaw.\n\nTwo smaller gaps: the covering comparison with the Biermé–Estrade model in Theorem 1.2 is plausible but abbreviated; a referee should ask for a more explicit argument that the high-frequency small balls cover the plane and hence the crosses do too. And the 'tedious but elementary calculation' in Theorem 1.3(ii) is really the core of the limit intensity; it should be written out or at least outlined with the key expansion steps.\n\nThe citation pattern is fine. The paper is clear and the thinking is serious throughout, especially the honest remark that connections to the L2 IPVT remain unclear.\n\nWho should read this: people working on low-intensity Voronoi limits, ideal tessellations, and random geometric structures on product spaces. It deserves serious refereeing, but the referee should insist on repairing the corona identification and expanding the abbreviated calculations before publication.","headline":"A genuine new IPVT construction on a non-hyperbolic L1 product space, with a real but fixable gap in the corona identification.","tokens_in":10971,"tokens_out":2214,"would_cite":true,"duration_ms":24168,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60D05","60G55"],"pacs":[],"model":"deepseek-v4-flash","headline":"Poisson–Voronoi limit yields an isometry-invariant tiling of H2×H2 in the L1 metric.","keywords":["ideal Poisson–Voronoi tessellation","product of hyperbolic planes","L1 metric","corona process","Gromov boundary","exponential separation","isometry invariance","ends of cells"],"falsifier":"Simulate a low-intensity Poisson point process on $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$ with intensity $\\lambda$, track all nuclei whose first coordinate remains within a fixed hyperbolic ball while the second coordinate diverges, and compute their renormalized delays $d(X_i, o) - \\log(1/\\lambda) + \\log\\log(1/\\lambda)$. If a non-zero fraction of these mixed-boundary nuclei have delays that do not diverge, the corona measure in the convergence theorem is incomplete and the resulting tessellation is not $\\mathrm{IPVT}(M)$.","tokens_in":9874,"feed_emoji":"🧩","tokens_out":8590,"duration_ms":72356,"temperature":0.7,"pith_summary":"This paper constructs the ideal Poisson–Voronoi tessellation of the product of two hyperbolic planes equipped with the $L^1$ metric, and proves that it is the low-intensity limit of ordinary Poisson–Voronoi tessellations. The limiting object is governed by a Poisson process on the corona $\\partial \\mathbb{H}^2 \\times \\partial \\mathbb{H}^2 \\times \\mathbb{R}_{\\ge 0}$, and its law is invariant under every isometry of the space. Almost surely, each cell reaches the boundary only in its end, the union of two circles determined by the cell's corona point, and the locus of points at equal separation from any two corona points is unbounded. The paper also identifies the stationary Poisson limit of the separation field seen from a point traveling toward a boundary intersection, including an exact tie-break probability. These results provide a concrete IPVT model in a space that is neither hyperbolic nor symmetric.","feed_headline":"Poisson–Voronoi limit yields an isometry-invariant tiling of H2×H2","feed_subtitle":"The zero-intensity limit is a corona process on the boundary; cells have explicit geometry and a tie-break probability of 0.70264.","key_machinery":"The mechanism is the corona process: a Poisson point process $(\\Theta_i, \\Phi_i, R_i)$ on $\\partial \\mathbb{H}^2 \\times \\partial \\mathbb{H}^2 \\times \\mathbb{R}_{\\ge 0}$ whose intensity is uniform on each boundary circle times Lebesgue on the radius coordinate. Together with the exponential separation formula $d(z,(\\theta,\\varphi,r)) = r\\, K(z_1,\\theta)K(z_2,\\varphi)$, this converts Voronoi comparisons into products of hyperbolic Poisson kernels. In the proof of the cell-end theorem, a horizontal plane is covered by algebraic ``hyperbolic crosses'' whose boundaries are the no-man's-lands between the zero cell and competing corona points; a Poisson random ball covering result shows these crosses cover the plane almost surely. The unbounded equal-separation claim follows from a zero-one lemma applied to the no-man's-land near a boundary-traveling point.","core_discovery":"The central claim is that when the intensity of a Poisson point process on $M = (\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$ tends to zero, the Voronoi diagram converges in law to a non-trivial random tessellation $\\mathrm{IPVT}(M)$ built from a Poisson process on the corona $\\partial \\mathbb{H}^2 \\times \\partial \\mathbb{H}^2 \\times \\mathbb{R}_{\\ge 0}$. Under this corona process, the separation from a point $z = (z_1, z_2)$ to a corona point $(\\theta, \\varphi, r)$ is exactly $r\\, K(z_1,\\theta) K(z_2,\\varphi)$, the product of two hyperbolic Poisson kernels. From this product formula the author derives isometry invariance of the tessellation, the almost-sure containment of each cell's boundary points in its end, unboundedness of the equal-separation locus, and an explicit stationary Poisson limit for the separation seen from a boundary-traveling observer, including the probability $1/2 + 2/\\pi^2$ that the mixed boundary point $(\\Theta_1, \\Phi_2)$ belongs to the zero cell.","pith_inferences":["Editorial extension: the same corona-product construction should extend verbatim to $m$-fold products $(\\mathbb{H}_{d_1} \\times \\cdots \\times \\mathbb{H}_{d_m}, L^1)$ with each $d_i \\ge 2$, since the Poincaré ball volume and Poisson kernel factor in each coordinate; the paper notes this possibility but proves only the case $m=2$, $d_i=2$.","Editorial extension: if the mixed-boundary nuclei with supposedly infinite delay were not negligible, the corona measure would not be the simple uniform-times-Lebesgue product; a numerical check of the renormalized delays of such nuclei at small intensity would separate the construction from a possible alternative corona carrying extra points.","Editorial extension: the isometry group of $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$ is the same as that of the Riemannian product $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^2)$, so the isometry-invariant corona measure constructed here may offer a bridge to the rank-two symmetric space tessellation studied in the fixed-price literature; whether the cell geometries are related is left open in the paper","Editorial extension: the exact tie-break probability $0.70264$ could be checked by direct Monte Carlo simulation of the low-intensity Poisson–Voronoi tessellation, giving a testable prediction of the limiting model."],"forward_implications":["The ideal tessellation is invariant under every isometry of $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$, so distributional properties computed for the zero cell transfer to all cells.","Almost surely, each cell meets the Gromov boundary only in its end, a union of two circles; no boundary point of a cell lies outside that end.","For any two corona points, the locus of points in $M$ at equal separation from them is unbounded almost surely, so no equality locus is confined to a compact region.","From the point of view of an observer traveling toward a boundary intersection, the rescaled separation field converges in law to a stationary Poisson process with an explicit product intensity measure on $\\mathbb{C} \\times \\mathbb{C} \\times \\mathbb{R}_{\\ge 0}$.","The probability that the mixed boundary point $(\\Theta_1, \\Phi_2)$ belongs to the zero cell is exactly $1/2 + 2/\\pi^2 \\approx 0.70264$, derived from a product of a uniform and two Beta$(1/2,1/2)$ variables."],"supporting_citations":[{"why":"Supplies the abstract deterministic convergence theorem and the Poisson-kernel separation framework that the paper applies to $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$.","marker":"[DCE+23]"},{"why":"Provides the analogous IPVT result for higher-rank symmetric spaces and motivates the equal-separation and separation-process statements proven here.","marker":"[FMW23]"},{"why":"Gives the Poisson random ball covering lemma used to show that hyperbolic crosses cover a plane almost surely in the proof of the cell-end theorem.","marker":"[BE12]"},{"why":"Contains the zero-one lemma used to upgrade local intersection with a no-man's-land into almost-sure unboundedness of the equal-separation set.","marker":"[Jeu82]"},{"why":"Supplies the isometry group of the Riemannian product $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^2)$, which the proof of the isometry-group lemma uses as a comparison.","marker":"[CT17]"},{"why":"Provides the transformation rule for hyperbolic Poisson kernels under Möbius maps needed for the limiting separation process.","marker":"[Bea12]"},{"why":"Defines the Gromov boundary and convergence used to place the vanishing-intensity nuclei at infinity.","marker":"[Gro81]"}],"fun_headline_variants":["Zero-intensity limit yields invariant tessellation on H2×H2","Corona process builds invariant Voronoi on H2×H2","Unbounded equal-separation sets in H2×H2 tessellation","Tie-break probability 0.70264 in Poisson-Voronoi"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction rests on the assertion that, in the vanishing-intensity limit, any nucleus converging to a mixed boundary point (one coordinate finite, the other at infinity) has infinite delay almost surely; if that assertion failed, the corona process would contain extra points and the separation formulas, isometry action, and cell-boundary descriptions would all have to change.","fun_headline_variants_meta":{"raw":{"variants":["Zero-intensity limit yields invariant tessellation on H2×H2","Corona process builds invariant Voronoi on H2×H2","Unbounded equal-separation sets in H2×H2 tessellation","Tie-break probability 0.70264 in Poisson-Voronoi"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000718,"raw_usage":{"total_tokens":3200,"prompt_tokens":897,"completion_tokens":2303,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":513,"completion_tokens_details":{"reasoning_tokens":2235}},"tokens_in":513,"tokens_out":2303,"duration_ms":16002,"temperature":1.0,"reasoning_tokens":2235,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:58:44.102865+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a low-intensity Poisson point process on $(\\mathbb{H}^2 \\times \\mathbb{H}^2, L^1)$ with intensity $\\lambda$, track all nuclei whose first coordinate remains within a fixed hyperbolic ball while the second coordinate diverges, and compute their renormalized delays $d(X_i, o) - \\log(1/\\lambda) + \\log\\log(1/\\lambda)$. If a non-zero fraction of these mixed-boundary nuclei have delays that do not diverge, the corona measure in the convergence theorem is incomplete and the resulting tessellation is not $\\mathrm{IPVT}(M)$.","supporting_citations":[],"review_version":1}