{"id":"09227825-bfef-4911-9184-afd5733cd2c3","arxiv_id":"2507.17037","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Barycentric discrete conformal maps between piecewise flat approximations of Riemannian surfaces converge, under fullness and local rigidity assumptions, to conformal maps, generalizing Rodin-Sullivan circle packing convergence.","lead":"This paper proves a general convergence theorem for discrete conformal maps on surfaces, showing that a wide family of piecewise flat approximations on curved surfaces have limits that are conformal (angle-preserving) maps. It unifies and extends the classic circle-packing convergence result of Rodin and Sullivan to many other discrete conformal structures.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"A hidden compatibility condition between geodesic edge lengths and discrete conformal factor edge lengths is not stated as a hypothesis, leaving Theorem 1 either vacuous for most geodesic triangulations or missing a necessary assumption.","rationale":"The reader's weakest assumption identifies the same compatibility issue, and it is indeed the most load-bearing concern. The theorem is conditional, so a missing hypothesis that restricts admissibility does not by itself make the theorem false; however, it determines whether the theorem has content for the claimed broad class of discrete conformal structures and geodesic triangulations. The paper never proves that a general geodesic triangulation can be realized as ℓ(f) in a fixed C_{α,η}, and the circle-packing example is the only place where compatibility is verified. Other concerns, such as the sketchiness of Lemma 25, the need for injectivity of the limit, and the verification of properness for vertex scaling, are real but appear more readily fixable within the existing framework. The concrete test with the pentagon shows that the compatibility condition is not vacuous and can fail for an otherwise ordinary geodesic triangulation, so it must be stated and handled explicitly. Since this concern supports the reader's CONDITIONAL verdict rather than overturning it, the verdict should remain unchanged.","tokens_in":30337,"tokens_out":25944,"duration_ms":334090,"concrete_test":"Take a triangulated disk obtained from a Euclidean pentagon with successive side lengths 1,2,1,2,1 by adding a central vertex connected to all boundary vertices with nondegenerate spoke lengths. For the circle-packing structure C_{1,1}, any solution must satisfy ℓ_ij = r_i + r_j. Around the boundary cycle, the alternating sum of the edge lengths is 1−2+1−2+1 = −1, but for any positive radii the alternating sum equals 2r_1 > 0, so no discrete conformal factor exists. Check whether this geodesic triangulation can appear in an admissible sequence: if the intended reading of Definition 32 requires ℓ(f_n) to equal the geodesic edge lengths, it cannot, exposing the missing compatibility hypothesis; if it does not, exhibit the two different metrics on Ω_n^Δ and verify that Corollary 15 cannot be applied to the metric used by Theorem 30.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Definition 16 constructs Ω_n^Δ from the geodesic edge lengths of a triangulated submanifold Ω_n. Definition 32 then requires the same Ω_n^Δ to equal (Ω_n, T_n, ℓ(f_n)) for a discrete conformal factor f_n in a fixed structure C_{α,η}. This equality is never stated explicitly, and it is not generally solvable. For the circle-packing structure C_{1,1}, edge lengths must satisfy ℓ_ij = e^{f_i} + e^{f_j}; around any even cycle this forces an alternating-sum condition on the edge lengths, and generic geodesic triangulations fail it. For vertex scaling C_{0,L^2/2}, one needs log(ℓ_ij/L_ij) to be a gradient on the graph, again a nontrivial cocycle condition. The proof of Proposition 34 depends on Corollary 15, which estimates Ψ^*g against the Euclidean metric g^Δ built from geodesic edge lengths, while Lemma 28 and Theorem 30 compare metrics built from ℓ(f_n) and ℓ(f̃_n). If the two sets of edge lengths differ, those estimates cannot be chained. Section 7 verifies compatibility only for the hexagonal circle-packing example; no general existence or construction is supplied. Thus, unless compatibility is added as an explicit admissibility condition, the theorem either has no content for most natural geodesic triangulations or has an unstated, load-bearing hypothesis.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a general convergence theorem for barycentric discrete conformal maps between Riemannian surfaces. Given an admissible sequence of geodesically triangulated submanifolds with the same combinatorics, discrete conformal factors in a fixed structure C_{α,η}, local discrete conformal rigidity (LDCR), fullness, and bounded factor ratios, the maps Φ_n = Ψ̃_n ∘ φ_n ∘ Ψ_n^{-1} are shown to have uniformly convergent subsequences; under an additional properness condition, the pullback metrics Φ_n^* g̃ converge in L∞ to eF g, so the limit is conformal. Section 7 applies the framework to recover the Rodin–Sullivan circle packing theorem, and Section 8 discusses applications to vertex scaling and other structures.","tokens_in":30551,"tokens_out":13188,"duration_ms":137930,"significance":"If the proof is completed, this would be a significant unification: one convergence framework covering circle packing, vertex scaling, and the general C_{α,η} structures, with or without boundary. The modular strategy is attractive and largely sound in outline: it combines the Riemannian barycentric estimates of [vDGW16] with new metric-distortion estimates for discrete conformal maps, and it isolates local discrete conformal rigidity as the key rigidity hypothesis. The paper is also honest about the role of fullness and explicitly uses the properness assumption only where needed. Several load-bearing points, however, are either unstated assumptions or insufficiently justified, most importantly the compatibility of the geodesic edge lengths with the edge lengths prescribed by the discrete conformal factors.","major_comments":[{"comment":"Definition 16 constructs Ω^Δ_n as the piecewise flat surface whose edge lengths are the geodesic edge lengths of the triangulated submanifold Ω_n. Definition 32, however, treats Ω^Δ_n as (Ω_n, T_n, ℓ(f_n)) for a conformal factor f_n in a fixed structure C_{α,η}, and the proof of Proposition 34 chains Corollary 15, which compares Ψ_n^* g with the PL metric built from geodesic edge lengths, with Theorem 30, which compares φ_n^* g̃^Δ with the PL metrics built from ℓ(f_n) and ℓ(f̃_n). This chaining is valid only if ℓ(f_n) and ℓ(f̃_n) coincide with the geodesic edge lengths of Ω_n and Ω̃_n. That equality is never stated as a hypothesis and is not automatic: for the circle packing structure C_{1,1}, the relation ℓ_{ij} = e^{f_i} + e^{f_j} is obstructed around even cycles by an alternating-sum condition, and for vertex scaling C_{0,L^2/2}, the relation log(ℓ_{ij}/L_{ij}) must satisfy a cocycle condition on every cycle. Section 7 verifies compatibility only for the hexagonal circle-packing example. The admissibility definition must either include this compatibility explicitly or the paper must prove it for the structures it claims to cover; otherwise Theorem 1 is vacuous for generic geodesic triangulations or silently depends on a strong unstated hypothesis.","section":"Definition 16 and Definition 32; Proposition 34"},{"comment":"Proposition 38 assumes a two-sided bound 1/H_K ≤ H_n(v) ≤ H_K on compact sets, but Definition 32, condition (2), only imposes a uniform upper bound on the ratio H_n(v) = e^{f̃_n(v)}/e^{f_n(v)}. The proof of Theorem 41 uses Proposition 38 and Corollary 40 to conclude that the limit eF is a positive continuous function. A uniform upper bound alone does not prevent liminf H_n from being zero on a compact set, in which case the claimed positive conformal factor would fail. Either add a uniform lower bound to the admissibility conditions or prove it from the other assumptions, since the positivity of eF is essential for the conformality conclusion in Theorem 41.","section":"Definition 32; Proposition 38; Corollary 40; Theorem 41"},{"comment":"The induction step in Lemma 25 asserts that if a shortest geodesic γ between ∂D_{m-1} and ∂D_m does not pass through a vertex, then the two boundary curves are parallel near the endpoints and γ can be translated until an endpoint is a vertex without changing its length. This is not justified on a piecewise flat surface with possible conical singularities: ∂D_m need not be a geodesic parallel curve, and no argument is given that such a translation remains inside the combinatorial disk or preserves length. Since Lemma 25 feeds into Lemma 27 and hence into Proposition 38 and Theorem 41, this step needs a rigorous proof or a citation. If the statement fails, the bound m ≥ R/(2ε_n) and the properness argument collapse.","section":"Lemma 25 (pages 12-13)"}],"minor_comments":[{"comment":"Definition 5 calls every piecewise linear map between two triangulated PL surfaces with the same combinatorics a discrete conformal map, with no condition relating the two discrete metrics. This conflicts with the usage in Definition 32 and with the metric estimates in Theorem 30; please add the missing discrete-conformal relation or rename the map.","section":"Definition 5"},{"comment":"The symbol Ω^Δ_n is overloaded: in Definition 16 it denotes the geodesic-length PL surface, while in Definition 32 and the surrounding text it is used for the ℓ(f_n)-PL surface. After the compatibility issue is resolved, please use distinct notation or state the equality explicitly.","section":"Notation throughout"},{"comment":"Condition (3) says the image set {Φ_n(x)} is contained in a compact subset V ⊂ N, but N is not defined in that context; the target manifold should presumably be M̃.","section":"Definition 32, condition (3)"},{"comment":"The proof asserts that each Φ_n is a homeomorphism on K without proving global injectivity; Lemma 19 gives only simplex-wise diffeomorphism, and Definition 16 only says that barycentric maps are assumed to exist. Please clarify why the global maps Ψ_n and Φ_n are homeomorphisms.","section":"Theorem 35 proof"}],"recommendation":"major_revision","confidential_remarks":"The claimed scope of the paper is broader than what is actually verified: the compatibility between geodesic edge lengths and edge lengths determined by a fixed discrete conformal structure is a serious missing hypothesis, and it affects the advertised applications to vertex scaling and general C_{α,η} structures as much as the main theorem. The modular estimates and the circle-packing recovery are valuable, so I would not reject the paper outright. The authors should be asked to add an explicit compatibility condition, prove a realization lemma for the structures they discuss, or narrow the theorem's scope; the lower-bound issue for H_n and the proof of Lemma 25 also need to be addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main theorem is the first convergence result covering the whole C_alpha,eta family of discrete conformal structures on Riemannian surfaces, with or without boundary. That is genuinely new: Rodin-Sullivan and the vertex scaling results become special cases. The pullback-metric proof via Riemannian barycentric coordinates is a real contribution, and making LDCR the explicit rigidity hypothesis is a useful conceptual step. Recovering Rodin-Sullivan in Section 7 is a good sanity check and shows the framework has content.\n\nThe soft spots are real but manageable. The biggest is the relation between the geodesic edge lengths of the triangulated submanifolds and the edge lengths ell(f_n) from the discrete conformal factors. Definition 16 builds Omega^Delta_n from geodesic lengths; Definition 32 then treats the same object as (Omega_n, T_n, ell(f_n)). For Proposition 34 to chain Corollary 15 with Theorem 30, those two edge-length assignments must coincide. The paper never states this compatibility as a hypothesis; it is implicit in 'discrete conformal factors for Omega^Delta_n', and the subscript slip in Definition 32 does not help. For vertex scaling any base triangulation works (take f=0); for circle packing only packable triangulations work, which Section 7 provides. So it is not fatal, but a referee should ask for the condition to be explicit and for a remark on when it holds.\n\nLemma 25's argument that a shortest geodesic between two combinatorial shells can be translated to a vertex without changing length is not justified. It is plausible, but as written it is hand-wavy. The rest of the proof outline is coherent. Theorem 35 gives equicontinuity and subsequential limits; Theorem 41 upgrades to conformality under properness. The limit is shown to be conformal, but injectivity and nonconstancy are not addressed; for the circle packing application they need the extra fact that the limit is a Riemann mapping. That is a gap in the application section, not in the main theorem.\n\nThe citation pattern is fine: substantial self-citation, but always to prior work they build on, not circular. The use of [vDGW16] and [GT17] as black boxes is appropriate.\n\nThis is for people working in discrete conformal geometry and circle packing; they will read it and use it. It deserves a serious referee. The main theorem is important enough, and the flaws are addressable. I would recommend sending it to review with a request to clarify the compatibility condition and tighten Lemma 25.","headline":"A genuinely unifying discrete conformal convergence theorem with a real but fixable compatibility issue and one sketchy lemma; worth serious refereeing.","tokens_in":31147,"tokens_out":5883,"would_cite":true,"duration_ms":57334,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53A70","52C26","65E10"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that barycentric discrete conformal maps converge, along subsequences, to conformal maps between Riemannian surfaces, with L∞ pullback-metric convergence under a properness assumption.","keywords":["discrete conformal mapping","circle packing","vertex scaling","Riemannian barycentric coordinates","piecewise flat surfaces","conformal convergence","Local Discrete Conformal Rigidity","discrete conformal structure"],"falsifier":"Compute, on one $(\\vartheta,\\epsilon)$-full triangle whose vertices are centers of combinatorial disks of generation $m$, both sides of the two-sided estimate $(1-\\beta\\epsilon)^2(1-C s_m)\\tilde F_n|X|^2_g \\le |X|^2_{\\Phi_n^*\\tilde g} \\le (1+\\beta\\epsilon)^2(1+C s_m)\\tilde F_n|X|^2_g$ from Proposition 34; an admissible sequence for which this estimate fails would falsify that proposition and hence Theorem 1.","tokens_in":30063,"feed_emoji":"📐","tokens_out":9863,"duration_ms":102540,"temperature":0.7,"pith_summary":"This paper proves a general convergence theorem for discrete conformal mappings between surfaces. It shows that if a sequence of triangulated approximations of two Riemannian surfaces is admissible, meaning the triangulations stay nondegenerate, the ratio of target to domain conformal factors is locally rigid in a precise sense, and a normalization point has compact image, then the associated barycentric discrete conformal maps have a subsequence converging uniformly on compact subsets to a continuous map. If the admissibility is proper, the pulled-back target metric converges in L∞ on compact subsets to a positive continuous function times the domain metric, which is exactly the statement that the limit is conformal. The result matters because it treats circle packing, vertex scaling, and the other discrete conformal structures of the same family in one framework, and it allows the domain and target to be curved surfaces with or without boundary, not just plane domains.","feed_headline":"Discrete conformal maps converge on any surface","feed_subtitle":"A general proof covers circle packing, vertex scaling, and curved targets with or without boundary.","key_machinery":"The load-bearing construction is the barycentric discrete conformal map $\\Phi_n = \\tilde\\Psi_n \\circ \\phi_n \\circ \\Psi_n^{-1}$, where $\\Psi_n$ and $\\tilde\\Psi_n$ are Riemannian barycentric (Karcher-mean) maps from piecewise-flat simplices to geodesic simplices of the two surfaces, and $\\phi_n$ is the piecewise-linear map between the piecewise-flat triangulations. The edge lengths on those triangulations are prescribed by a discrete conformal structure $C_{\\alpha,\\eta}$ through $\\ell_{ij}^2 = \\alpha_i e^{2f_i} + \\alpha_j e^{2f_j} + 2\\eta_{ij} e^{f_i+f_j}$. The argument has three compression stages: the barycentric distortion estimates compare the pullback of the smooth metric to the flat metric; an edge-length comparison shows the flat metric in the image is close to a conformally rescaled flat metric in the domain; and the Local Discrete Conformal Rigidity condition makes the ratio of target-to-domain conformal factors nearly constant over combinatorial disks of generation $m$, with a control sequence $s_m \\to 0$. Fullness of simplices keeps triangles from degenerating, and properness, meaning $s_m \\le \\alpha/m$, is what upgrades uniform convergence of maps to $L^\\infty$ convergence of pulled-back metrics.","core_discovery":"The central discovery is that convergence to conformal maps is a general phenomenon for discrete conformal structures, not a special feature of circle packings. For any discrete conformal structure in the family C_{\\$\\alpha$,\\eta} (edge lengths of the form $\\ell_{ij}^2 = \\alpha_i e^{2f_i} + \\alpha_j e^{2f_j} + 2\\eta_{ij} e^{f_i+f_j}$) and any pair of surfaces with generalized triangulated exhaustions satisfying the admissibility hypotheses, the barycentric discrete conformal maps $\\Phi_n = \\tilde\\Psi_n \\circ \\phi_n \\circ \\Psi_n^{-1}$ admit a subsequence converging uniformly on compact sets. When the sequence is proper, the pullback metrics converge: $\\Phi_n^* \\tilde g \\to \\tilde F g$ in $L^\\infty$ on compact subsets, with $\\tilde F$ positive and continuous, which implies the limiting map is conformal. The proof obtains this by chaining two Riemannian barycentric distortion estimates around a piecewise-linear discrete conformal map and controlling the discrete conformal factors with the Local Discrete Conformal Rigidity condition.","pith_inferences":["A reader applying this theorem to a new discrete conformal structure should check that the geodesic edge lengths of each approximating triangulation are realizable as $\\ell(f_n)$ for some conformal factor $f_n$; the paper itself verifies this compatibility only for the circle-packing example.","The structure of the proof suggests a natural experiment: if the LDCR constants $s_m$ decay slower than $1/m$, map convergence may still hold through Theorem 35 while the $L^\\infty$ metric convergence of Theorem 41 may fail, isolating exactly what the properness assumption buys.","Because the convergence is subsequential, uniqueness of the limit is not addressed; combining the theorem with a three-point or boundary normalization would be the natural next step toward a full discrete uniformization statement.","The $L^\\infty$ convergence of pulled-back metrics implies convergence of the discrete conformal factors as functions, which could be used in future work to prove convergence of discrete curvatures or discrete energies under the same hypotheses."],"forward_implications":["The classical circle-packing convergence theorem is recovered as a special case: for simply connected bounded plane domains, the hexagonal circle-packing construction of Section 7 yields a proper admissible sequence, so the theorem gives subsequential convergence to a Riemann mapping.","For vertex-scaling structures, sequences built from a prescribed conformal map, as in the construction discussed in Section 8, satisfy the admissibility hypotheses, so the same theorem covers those convergence results without a separate quasiconformality argument.","Because every discrete conformal structure in the family is a $C_{\\alpha,\\eta}$ structure, the theorem applies uniformly to circle packing, vertex scaling, and related structures; the user only needs to verify exhaustions, fullness, LDCR, and the ratio bound.","The properness condition provides a quantitative conclusion: the pulled-back target metric converges in $L^\\infty$ on compact subsets to $\\tilde F g$, meaning the discrete conformal factors themselves converge to a continuous function on the limit domain.","The theorem allows both the domain and target to be curved Riemannian surfaces with or without boundary, extending earlier convergence results that were restricted to plane domains or flat tori."],"supporting_citations":[{"why":"Supplies the Riemannian barycentric distortion estimates that compare smooth and piecewise-flat pullback metrics, the central comparison used in Proposition 34.","marker":"[vDGW16]"},{"why":"Provides the original circle-packing convergence theorem and the hexagonal-packing and length-area estimates used to build the Section 7 admissible sequence.","marker":"[RS87]"},{"why":"Shows that the discrete conformal structures considered all have the $C_{\\alpha,\\eta}$ form, so one framework covers them.","marker":"[GT17]"},{"why":"Supplies the bounded-valence generalization of the hexagonal packing lemma, with the rate estimate that makes circle-packing sequences proper.","marker":"[HR93]"},{"why":"Supplies Schwarz-lemma-type ratio bounds for circle packings used to bound the ratio of conformal factors in the circle-packing admissible sequence.","marker":"[Rod87]"},{"why":"Provides the vertex-scaling convergence theorem and the rigidity lemma used to verify the hypotheses in the vertex-scaling setting.","marker":"[LSW22]"},{"why":"Establishes existence and uniqueness of the Karcher mean used to define the Riemannian barycentric maps.","marker":"[Kar77]"}],"fun_headline_variants":["Universal convergence of discrete conformal maps","Circle packing no longer special: discrete conformal convergence","New theorem: discrete conformal maps converge everywhere","Conformal convergence unifies discrete structures"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the geodesic edge lengths of each triangulated submanifold coincide with the edge lengths produced by the chosen discrete conformal factors through the formula $C_{\\alpha,\\eta}$; if a geodesic triangulation is not realizable in that way, the discrete conformal map and the barycentric comparison cannot both hold, and the theorem has no content for that sequence.","fun_headline_variants_meta":{"raw":{"variants":["Universal convergence of discrete conformal maps","Circle packing no longer special: discrete conformal convergence","New theorem: discrete conformal maps converge everywhere","Conformal convergence unifies discrete structures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000208,"raw_usage":{"total_tokens":1408,"prompt_tokens":952,"completion_tokens":456,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":400}},"tokens_in":568,"tokens_out":456,"duration_ms":5758,"temperature":1.0,"reasoning_tokens":400,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:59:40.799969+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, on one $(\\vartheta,\\epsilon)$-full triangle whose vertices are centers of combinatorial disks of generation $m$, both sides of the two-sided estimate $(1-\\beta\\epsilon)^2(1-C s_m)\\tilde F_n|X|^2_g \\le |X|^2_{\\Phi_n^*\\tilde g} \\le (1+\\beta\\epsilon)^2(1+C s_m)\\tilde F_n|X|^2_g$ from Proposition 34; an admissible sequence for which this estimate fails would falsify that proposition and hence Theorem 1.","supporting_citations":[],"review_version":1}