{"id":"73d2a00b-652c-4330-b7b9-75a0b1595aef","arxiv_id":"2501.01678","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The combinatorial Calabi flow for ideal circle patterns exists for all time and converges exponentially fast to the target ideal circle pattern metric whenever the target curvature vector is attainable.","lead":"This paper studies a combinatorial Calabi flow that gradually adjusts the radii of circles in a pattern on a surface, and proves the flow always runs to completion and settles exponentially fast at a pattern with any prescribed curvatures that are mathematically allowed. This gives a new algorithm for constructing ideal circle patterns, objects linked to hyperbolic geometry and 3-manifold topology.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the proof is internally consistent given the cited estimates, and the reliance on Lemma 5 is a normal external dependency rather than a detected flaw.","rationale":"I read the paper in good faith and checked the chain of implications leading to Theorem 1. The hyperbolic proof transforms the original flow into the u-coordinate flow, defines a strictly convex Lyapunov function, uses positivity of the Jacobian L to obtain monotonicity, and then applies Lemma 5 to prevent r_i from escaping to +infinity. The Euclidean proof is parallel but relies on coercivity on the invariant hyperplane, not on Lemma 5. I re-derived the key inequality in the proof of Theorem 2 and confirmed that Lemma 5, as stated, supplies exactly the needed estimate after choosing the appropriate rho. I also considered the possible concern that Lemma 7(ii) might fail for unbounded convex domains, but along any ray from the critical point the one-dimensional restriction has positive second derivative and zero derivative at the critical point, so the function grows at least linearly; hence the cited coercivity lemma is plausible and standard. The proof of exponential convergence is a standard Lyapunov/ODE argument and is internally coherent. The only genuine dependency is on external results, especially Lemma 5 and Theorem A, which are not reproved. This is a normal feature of research papers and not, by itself, a defect. No internal inconsistency or unstated assumption that would break the central claim was found. The verdict should remain unchanged, with the same caveat about reliance on imported technical lemmas that the reader already recorded.","tokens_in":8311,"tokens_out":37896,"duration_ms":357226,"concrete_test":"Independently verify Lemma 5 for ideal hyperbolic circle patterns satisfying (C1): for a concrete triangulation (e.g., the four vertices of a tetrahedron with a prescribed angle function), compute the asymptotic behavior of dK_j/du_i and dK_i/du_i as r_i goes to infinity, and confirm that for every rho > 1 the combination rho * sum_{j~i} dK_j/du_i + (rho - 1) * dK_i/du_i is eventually positive, uniformly in the other radii. If this inequality fails in any such example, the hyperbolic long-time existence proof collapses; if it holds, the main theorem stands.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. For Theorem 1 to hold, the hyperbolic escape-to-infinity argument in the proof of Theorem 2 must be valid; the only step that rules out r_i going to +infinity is Lemma 5, whose proof is imported from [8, 11, 13, 23] and not reproduced here. I checked the use of Lemma 5 in Section 3.2 and found it arithmetically consistent: with A = -sum_{j~i} dK_j/du_i > 0 and B = dK_i/du_i > 0, choosing rho = (2*pi - kmin)/(2*pi*eta - kmax) > 1 converts the lemma into B/A > rho, which is exactly the positivity needed to make du_i/dt < 0 for large r_i. Lemma 4 and Lemma 7, also cited, are standard and are applied correctly. The Euclidean case avoids Lemma 5 entirely and uses the coercivity of the Lyapunov function on the invariant hyperplane. Thus, assuming the cited lemmas are correct, the central argument of the paper is sound. The remaining risk is the correctness of the imported technical estimates, especially Lemma 5, not an internal inconsistency in the paper.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies combinatorial Calabi flow for ideal circle patterns on closed triangulated surfaces, in both hyperbolic and Euclidean background geometry. The flow is defined in terms of discrete curvature K and a prescribed attainable target curvature vector k. The main result, Theorem 1, asserts that under the angle condition (C1) and attainability of k, the flow exists for all time and converges exponentially fast to a radius vector whose associated ideal circle pattern metric has discrete curvature k. The proofs go through a change of variables u (u_i = ln tanh(r_i/2) in the hyperbolic case, u_i = ln r_i in the Euclidean case), a Lyapunov function Lambda defined as the integral of (K-k)·du, strict convexity and coercivity of Lambda, and an energy-decay argument for exponential convergence. In the hyperbolic case, long-time existence is handled separately: boundary behavior toward u_i -> -infinity is controlled by the coercivity of Lambda, while escape to +infinity is ruled out using a sign estimate (Lemma 5 imported from prior work by Ge-Hua, Glickenstein-Thomas, Wu-Xu, and Li-Luo-Xu). The Euclidean case avoids the +infinity issue by working on the invariant hyperplane sum_i u_i = constant and using the coercivity of Lambda there.","tokens_in":8522,"tokens_out":1649,"duration_ms":19146,"significance":"If the result holds, it is a clean and useful extension of the combinatorial Calabi flow program to ideal circle patterns, providing a variational method and an algorithmic route to construct ideal circle pattern metrics with prescribed discrete curvature. The paper is concise, and the central geometric ingredients (strict convexity of the energy, positivity of the Jacobian, exponential convergence via the energy C) are standard and correctly assembled. The paper also gives explicit credit to the prior work on which it relies, especially Lemma 5, and it records the independent related work of X. Zhang. The main mathematical risk is the correctness of the imported Lemma 5 for ideal circle patterns in the hyperbolic regime; the paper does not prove Lemma 5, so the long-time existence proof in Theorem 2 is conditional on an external estimate. That is a normal dependency in the field, not an internal flaw. Overall the paper is a solid contribution that fits the journal's scope.","major_comments":[{"comment":"The long-time existence proof for the hyperbolic flow depends entirely on Lemma 5 to rule out ri(t) -> +infinity. Lemma 5 is not proved in the paper; it is imported from [8], [11], [13], [23]. The use of the lemma is arithmetically consistent: with A = -sum_{j~i} dK_j/du_i > 0 and B = dK_i/du_i > 0, the condition rho*A - (rho-1)*B > 0 is exactly the positivity required in the display after (3.2). However, since Theorem 2 is the only place where escape to +infinity is excluded for the hyperbolic flow, the main theorem for hyperbolic background geometry is conditional on the correctness of that external estimate. I would like the authors to state this dependency explicitly in the text and, if possible, to include a proof or a precise reference to the exact statement of Lemma 5 in each cited source. This is a load-bearing external dependency, not a fatal flaw, but it should be transparently acknowledged.","section":"Section 3.2, Theorem 2"},{"comment":"In the proof of exponential convergence, after deriving |Ki - ki| <= sqrt(C(u(0))) e^{-lambda0 t}, the authors estimate |ui - u*_i| by integrating d(ui - u*_i)/dt = sum_j (Kj - kj) dKj/du_i. The bound uses uniform boundedness of dKj/du_i on the compact set containing u(t). This step is correct, but the constant lambda in the final display is not explicitly related to lambda0 and the uniform bound; it would be helpful to spell out how the two constants combine (e.g., lambda = sqrt(C(u(0))) * sup |dKj/du_i| / lambda0). The claim as written is acceptable, but the estimate deserves a one-line justification.","section":"Section 3.3, Theorem 3"},{"comment":"The proof of long-time existence in the Euclidean case avoids Lemma 5 entirely and uses the coercivity of Lambda on the invariant hyperplane. However, the claim that 'the flow never touches the boundary of Υ in any finite time interval' is not fully justified as written. Since u stays in a compact set of Υ whenever T0 is finite (by the contradiction argument), the statement follows, but the wording is loose. Also, the proof of the 'similar argument' for exponential convergence is not given; given that the Jacobian is only positive definite on Υ (with null space along (1,...,1)), it is worth noting explicitly that the energy C decays via L^2 restricted to Υ and that the same computation as in Theorem 3 applies. This is a minor expositional gap, not an error.","section":"Section 4.2, Theorem 4"}],"minor_comments":[{"comment":"The notation T h is used both for the curvature map and for the hyperbolic background geometry; the superscript is h, which is fine, but the map from R^N_+ to R^N and the map from R^N_- (or R^N) in u-coordinates could be given different names or a remark that they are the same map in different coordinates.","section":"Section 1.2"},{"comment":"Lemma 5 is stated without specifying the precise quantitative dependence of M on rho and on the triangulation/angle function. Since the lemma plays a central role, a phrase such as 'for some M = M(rho, T, Theta, ...)' would clarify the uniformity in the subsequent argument.","section":"Section 2, Lemma 5"},{"comment":"The condition Ki > 2πη > kmax is used, but the parameter η is introduced via Lemma 3. It would be clearer to define η as a fixed number in (0,1) such that kmax < 2πη, and to state explicitly that Lemma 3 gives M1 depending on η. The current text is correct but a bit implicit.","section":"Section 3.2, equation (3.2)"},{"comment":"The proof of Lemma 8 is omitted with a reference to the hyperbolic argument. Since the Euclidean domain is R^N rather than R^N_-, and the invariance of sum_i u_i is used to restrict to Υ, a sentence explaining that Lemma 7 still applies on the unbounded convex set Υ would improve readability.","section":"Section 4.1"},{"comment":"The paper has several typographical issues: 'P A TTERN' in the title header, 'W ang' for the author name, and a space before the comma in 'j ∼ i' in the display after (3.3). These should be corrected in the final version.","section":"General"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a clean, short contribution that mostly assembles existing machinery. The strongest reason for minor revision rather than accept is the external dependency on Lemma 5 in the hyperbolic long-time existence proof; the authors should either quote the lemma with full precision or relegate it to a clearly marked assumption, so that readers and referees know exactly what is being imported. There is no detected internal inconsistency. The 'Statement of independent research outcomes' is unusual but not problematic; it does not affect the mathematical assessment. The manuscript is appropriate for the journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the bottom line: this paper proves a genuinely new convergence theorem for the combinatorial Calabi flow on ideal circle patterns, in both hyperbolic and Euclidean background geometry. Prior Calabi flow results (Ge-Hua) handled circle patterns with non-obtuse exterior angles; prior Ricci flow work (Ge-Hua-Zhou) handled ideal patterns but with a different flow. This paper gets the Calabi flow for the ideal case, which requires controlling radii near 0 and infinity. The main theorem is clean and the proof is logically solid.\n\nWhat the paper does well: it constructs the right Lyapunov function, shows it is strictly convex and proper on the domain, and uses it to rule out boundary blow-up in finite time. The hyperbolic escape-to-infinity argument is the delicate part; the Euclidean case is handled more simply using the invariant hyperplane. The paper is honest about its sources and even notes a possible concurrent independent result. That is good scholarly practice.\n\nSoft spots, in proportion: the biggest dependency is Lemma 5, imported from Ge-Hua, Glickenstein-Thomas, Wu-Xu, and Li-Luo-Xu. The paper does not prove it and gives no sketch. This lemma is the essential tool for preventing a radius from escaping to +infinity in the hyperbolic flow. I checked the application of the lemma and it is arithmetically correct: a standard choice of the parameter converts the lemma's inequality into exactly the positivity the proof needs. So the dependency is real but legitimate. A referee would still want the lemma stated in a self-contained way and the reliance flagged. The Euclidean part is terse—\"by a similar argument\"—and the final convergence proof there is mostly sketched. For an expert this is fine; for verification it is a bit terse.\n\nThe citation pattern is appropriate; the self-citations are to closely related technical results. No circularity: the target curvature vector K is an input, not derived from the flow.\n\nWho is this for: researchers working on combinatorial flows, circle patterns, and discrete conformal geometry. It will be a useful result to have in the literature. My recommendation: send it to a serious referee. It is not a result that will reshape the field, but it is correct, new, and well-framed.","headline":"New and correct convergence result for combinatorial Calabi flow on ideal circle patterns, with one imported technical lemma as the main dependency; deserves refereeing.","tokens_in":9065,"tokens_out":10462,"would_cite":true,"duration_ms":84536,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52C26"],"pacs":[],"model":"deepseek-v4-flash","headline":"The combinatorial Calabi flow exists for all time and converges exponentially fast to ideal circle patterns with prescribed discrete curvatures.","keywords":["ideal circle pattern","combinatorial Calabi flow","discrete curvature","hyperbolic background geometry","Euclidean background geometry","exponential convergence","curvature map","circle patterns"],"falsifier":"On a genus-2 surface with any triangulation, choose an angle function $\\Theta$ satisfying (C1) and a hyperbolic attainable $K$, then integrate (1.4); if the solution escapes to $0$ or $+\\infty$ in finite time or does not converge exponentially, Theorem 1 is false. A cheaper algebraic check is to evaluate the Lemma 5 quantity $\\rho\\sum_{j\\sim i}\\partial K_j/\\partial u_i+(\\rho-1)\\partial K_i/\\partial u_i$ for large $r_i$ on that triangulation and see whether it is eventually positive for some $\\rho>1$.","tokens_in":8097,"feed_emoji":"⭕","tokens_out":7661,"duration_ms":70336,"temperature":0.7,"pith_summary":"The paper proves that the combinatorial Calabi flow, a curvature-decreasing evolution for the radii of circles in a triangulated surface, works for ideal circle patterns in both hyperbolic and Euclidean background geometry. Under the angle condition (C1) and the assumption that the target curvature vector is attainable, the flow exists for all time and converges exponentially fast to a radius vector realizing the target curvatures. This turns the static existence problem for ideal circle patterns into a concrete algorithm: run the flow and read off the radii. The proof combines strict convexity of a well-defined potential with a large-radius derivative estimate, plus exponential decay of a squared-curvature energy.","feed_headline":"Calabi flow converges to ideal circle patterns","feed_subtitle":"A curvature-decreasing flow on surfaces provably runs forever and hits target circle-pattern curvatures.","key_machinery":"The engine is the curvature map written in the logarithmic variables $u_i=\\ln\\tanh(r_i/2)$ (hyperbolic) or $u_i=\\ln r_i$ (Euclidean), together with the potential $\\Psi(u)=\\int_{u(0)}^u\\sum_i(K_i-k_i)\\,du_i$, whose Hessian is the Jacobian matrix $L$ of the curvature map. Strict convexity of $\\Psi$ (from Lemma 4, where $L$ is positive definite, on a hyperplane in the Euclidean case) gives a unique critical point, the target radius vector. Long-time existence uses Lemma 5, an imported estimate that for any $\\rho>1$ there is an $M$ such that $\\rho\\sum_{j\\sim i}\\partial K_j/\\partial u_i+(\\rho-1)\\partial K_i/\\partial u_i>0$ whenever $r_i>M$, to rule out a radius escaping to $+\\infty$; coercivity of $\\Psi$ rules out escape to $0$. Exponential convergence comes from the energy $C(u)=\\sum_i(K_i-k_i)^2$, whose time derivative is $-2(K-k)^T L^2(K-k)\\le -2\\lambda_0 C(u)$.","core_discovery":"On the paper's own terms, Theorem 1 is the central discovery: for any angle function $\\Theta\\colon E\\to(0,\\pi)$ satisfying (C1), and any hyperbolic (resp. Euclidean) attainable curvature vector $K=(k_1,\\dots,k_N)$, the flow (1.4) (resp. (1.5)) has a unique solution for all time and converges exponentially fast to a radius vector $r^*$ such that the ideal circle pattern metric it determines has discrete curvatures $K$. In hyperbolic background this target is unique; in Euclidean background it is unique up to scaling, and the flow preserves the normalization $\\sum_i u_i$ constant. The convergence proof bounds the squared curvature error by $C(u(0))e^{-2\\lambda_0 t}$ on the compact invariant set where the flow lives, then transfers this exponential decay to the radius variables.","pith_inferences":["Since the proof treats Lemma 5 as a black box, a natural extension is to check numerically on random triangulations whether the Lemma 5 positivity holds for ideal circle patterns; if it ever fails, the hyperbolic long-time existence argument would need a different cap on radii.","The same Lyapunov-plus-coercivity template should apply to other combinatorial curvature flows for ideal patterns, such as $p$-th Calabi flows, provided an analogous large-radius estimate can be proved.","Quantifying $\\lambda_0$, the smallest eigenvalue of $L^2$ on the invariant compact set, would turn the exponential convergence statement into an explicit rate and make the algorithm practically predictable.","Because ideal circle patterns correspond to ideal hyperbolic polyhedra, this flow can be read as a discrete Calabi deformation driving an ideal polyhedron to prescribed dihedral angles, complementing Rivin's uniqueness theorem."],"forward_implications":["For every curvature vector that satisfies the Bobenko–Springborn inequalities, the flow provides a constructive way to realize an ideal circle pattern metric with exactly those curvatures.","Because the convergence rate is exponential, discretized versions of (1.4) and (1.5) give a practical numerical algorithm for finding ideal circle patterns, not just an existence proof.","In the Euclidean case the flow automatically fixes the scaling ambiguity by preserving $\\sum_i u_i$, so the limiting pattern is determined up to the same homothety freedom as the static problem.","The theorem extends the combinatorial Calabi flow from the non-obtuse-circle-pattern setting of earlier work to ideal circle patterns, where only the angle sum condition (C1) is required.","After the change of variables, both the hyperbolic and Euclidean flows take the same gradient form $\\dot u=-(K-k)^T L$ in the appropriate domain, so the two cases are unified."],"supporting_citations":[{"why":"Supplies Theorem A, the complete characterization of the curvature image and the existence of a radius vector for every attainable curvature vector.","marker":"[2]"},{"why":"Supplies Lemmas 1–4, including the angle derivative signs and positive-definite Jacobian for ideal circle patterns used throughout the proof.","marker":"[10]"},{"why":"Introduced the hyperbolic combinatorial Calabi flow framework and gave the initial large-radius derivative estimate underlying Lemma 5.","marker":"[8]"},{"why":"Provides the duality-structure result used in one proof of the key large-radius estimate in Lemma 5.","marker":"[11]"},{"why":"Gives an independent proof of Lemma 5, the estimate the hyperbolic long-time existence argument relies on.","marker":"[23]"},{"why":"Gives another independent proof of Lemma 5 and earlier combinatorial Calabi flow results that the paper extends.","marker":"[13]"},{"why":"Introduced combinatorial Ricci flows on surfaces and the variational 1-form technique that the potential function here follows.","marker":"[3]"},{"why":"Introduced combinatorial Calabi flows on surfaces, the direct predecessor of the flow studied in this paper.","marker":"[7]"}],"fun_headline_variants":["Calabi flow never times out, converges to circle patterns","Exponential convergence: Calabi flow solves circle patterns","Ideal circle patterns: Calabi flow proven to converge","Flow exists for all time, meets target curvatures"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Lemma 5, an imported estimate saying that once one vertex radius is large enough, a particular weighted sum of curvature derivatives at that vertex is strictly positive; the paper does not prove this estimate for ideal circle patterns, and the hyperbolic long-time existence proof relies on it.","fun_headline_variants_meta":{"raw":{"variants":["Calabi flow never times out, converges to circle patterns","Exponential convergence: Calabi flow solves circle patterns","Ideal circle patterns: Calabi flow proven to converge","Flow exists for all time, meets target curvatures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000215,"raw_usage":{"total_tokens":1329,"prompt_tokens":746,"completion_tokens":583,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":362,"completion_tokens_details":{"reasoning_tokens":530}},"tokens_in":362,"tokens_out":583,"duration_ms":6464,"temperature":1.0,"reasoning_tokens":530,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:23:01.011154+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a genus-2 surface with any triangulation, choose an angle function $\\Theta$ satisfying (C1) and a hyperbolic attainable $K$, then integrate (1.4); if the solution escapes to $0$ or $+\\infty$ in finite time or does not converge exponentially, Theorem 1 is false. A cheaper algebraic check is to evaluate the Lemma 5 quantity $\\rho\\sum_{j\\sim i}\\partial K_j/\\partial u_i+(\\rho-1)\\partial K_i/\\partial u_i$ for large $r_i$ on that triangulation and see whether it is eventually positive for some $\\rho>1$.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Theorem A, the complete characterization of the curvature image and the existence of a radius vector for every attainable curvature vector."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Lemmas 1–4, including the angle derivative signs and positive-definite Jacobian for ideal circle patterns used throughout the proof."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced the hyperbolic combinatorial Calabi flow framework and gave the initial large-radius derivative estimate underlying Lemma 5."},{"cited_title":"Glickenstein, J","cited_arxiv_id":null,"evidence_quote":"Provides the duality-structure result used in one proof of the key large-radius estimate in Lemma 5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives another independent proof of Lemma 5 and earlier combinatorial Calabi flow results that the paper extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced combinatorial Ricci flows on surfaces and the variational 1-form technique that the potential function here follows."},{"cited_title":"Ge, Combinatorial Calabi flows on surfaces , Trans","cited_arxiv_id":null,"evidence_quote":"Introduced combinatorial Calabi flows on surfaces, the direct predecessor of the flow studied in this paper."}],"review_version":1}