{"id":"944f820e-2706-42f4-9eb3-ae514a3530fb","arxiv_id":"1908.09733","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For families of log curves carrying a mesa structure, the selected subcurve can be contracted in a base-change-compatible way, yielding contractions between moduli spaces of curves.","lead":"This paper defines a class of logarithmic curves called mesa curves and proves that the chosen subcurve can always be contracted within every family, without changing genus and compatibly with base change. The result gives a general machine for producing maps between moduli spaces of curves, with singularities that include the elliptic Gorenstein ones.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Acyclicity H^1(E,O_E(-λ))=0 is the true boundary of the construction; Section 1.1 concedes it fails for genus-two Gorenstein models, so the theorem is narrower than the introduction may suggest.","rationale":"I reviewed the proof for internal gaps in the main steps: Lemma 4.15 is a standard truncation statement whose omitted proof is routine; Theorem 4.16's base-change step is justified by the vanishing of H^{i+1} and flatness of the Cech terms, even though the 'right exact' wording is terse; the colimit identification in Proposition 4.18 is plausible in the relative-curve setting; and Proposition 4.25's independence-of-sections argument is sketchy but appears standard. The one hypothesis that genuinely cannot be relaxed is the acyclicity H^1(E,O_E(-λ))=0: it is what makes the cohomology vanish in Proposition 4.18 and hence what makes the contracted curve flat and base-change compatible. The author explicitly acknowledges in Section 1.1 that this fails for the genus-two Gorenstein models in Battistella's work, so the main theorem does not deliver those contractions. That is a real limitation of the central claim's reach, but it is an honest modeling condition rather than a hidden assumption or a proof error. The reader's weakest-assumption analysis identified exactly this point, and my read does not change the verdict.","tokens_in":37502,"tokens_out":40023,"duration_ms":411434,"concrete_test":"Take a semistable model of Battistella's genus-two Gorenstein singularity, with the divisor λ as in Section 1.1, and compute H^1(E,O_E(-λ)) directly on the singular fiber. If it is nonzero as claimed, then run the §4 construction without imposing acyclicity: check whether the B-sequence remains short exact and whether U is flat over S. If flatness or base-change compatibility fails in this example, the acyclicity hypothesis is essential and the theorem's scope is exactly as bounded; if it still holds, the condition could be relaxed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central theorem is explicitly conditional on the mesa-curve hypotheses, and among those the acyclicity condition H^1(E,O_E(-λ))=0 is the load-bearing one. It is used in Proposition 4.18 to obtain the vanishing of H^q(U,O_U(-λ)) for q≥1, which in turn underlies the flatness of U over S, the short exactness of the B-sequence in Proposition 4.19, and the base-change compatibility asserted in Theorem 1.1(ii). Without acyclicity, the H^1(E_t,O_{E_t}(-λ)) term in the proof of Proposition 4.18 cannot be killed by enlarging nΣ, because the sections Σ are chosen to avoid E; the argument collapses exactly at that point. The same condition is also the criterion in Lemma 4.14 that lets one recognize admissible inputs. Section 1.1 states that the topological models of Battistella's genus-two Gorenstein singularities fail this condition, so the construction does not produce the genus-two contractions one might expect from the introduction's framing. This is not an internal inconsistency: the theorem is a theorem about acyclic mesa curves, and the author is transparent about the boundary. The concern is scope, not soundness.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the notion of a mesa curve—a log curve equipped with a section λ of the characteristic sheaf whose associated piecewise linear function has a 'mesa' shape—and constructs, for each family of mesa curves, a contraction of the support E=|λ|. The contraction is built by explicitly defining a structure sheaf: locally one takes Spec of a ring B(U) built from Γ(U,O_C(−λ)) and Γ(S,O_S), then glues it to C−E via a pushout. The main theorem (Theorem 1.1) asserts that this contraction is a family of curves, commutes with arbitrary base change, and that steep genus-one components are contracted to elliptic Gorenstein singularities. The proof is divided into reduction to a standard situation, vanishing and flatness via a cohomology-and-base-change argument, well-definedness and independence of auxiliary sections, and an explicit description of the singularity in §4.6. Section 1.1 explicitly notes that the construction does not cover Battistella's genus-two Gorenstein models because they fail the acyclicity condition H^1(E,O_E(−λ))=0.","tokens_in":37774,"tokens_out":33111,"duration_ms":342786,"significance":"If correct, the theorem provides a base-change-compatible contraction for a broad class of subcurves of log curves with explicit control of the resulting singularities, generalizing the genus-one constructions of Ranganathan–Santos-Parker–Wise and Parker and giving a potential route to modular compactifications of M_{g,n} and to desingularizations of stable maps. The paper's method—constructing the structure sheaf directly rather than via a relative Proj—is a useful contribution, and the singularity analysis in §4.6 is concrete and informative. The main result is conditional on the mesa axioms, and the author is transparent about the acyclicity boundary. The proof is long but mostly careful; however, the validity of one of its cornerstones, Lemma 4.15, is doubtful and needs repair before the flatness and base-change claims can be accepted.","major_comments":[{"comment":"Lemma 4.15 is stated without proof and is false as stated. Take A=k[ε]/(ε^2) and the cochain complex C• with C^i=A for i≥0 and differentials all equal to multiplication by ε. Each C^i is flat, and H^0(C•)=εA≅k is not flat over A, while H^i(C•)=0 for all i≥1; thus the hypotheses hold with j=0. The usual smart truncation τ≤0C• has degree-0 term ker d^0=εA, which is not flat, and the naive truncation is not quasi-isomorphic to C•. Consequently the assertion in Lemma 4.15 that τ≤jC• is a complex of flat A-modules fails, and the proof of Theorem 4.16, which uses Lemma 4.15 to obtain both base-change compatibility and flatness of H^0, is invalid. Since Theorem 4.16 is used in Proposition 4.18 and again in Proposition 4.19 to establish exactness, flatness, and base change for the B-sequence, this gap is load-bearing for Theorem 1.1. The intended results may be true—they are standard consequences of the cohomology-and-base-change theorem—but the paper needs either a correct replacement lemma with proof or a direct citation of the standard theorem.","section":"§4.4, Lemma 4.15 and Theorem 4.16"}],"minor_comments":[{"comment":"The abstract describes contractions for curves 'of any genus' without mentioning the acyclicity hypothesis H^1(E,O_E(−λ))=0 that is built into Definition 3.2(vii). Because §1.1 explicitly states that this condition fails for the genus-two Gorenstein models of [Bat19], the abstract should state that the theorem concerns acyclic mesa curves, so that readers are not led to expect those contractions.","section":"Abstract and §1.1"},{"comment":"The sentence 'Since π:E→S is proper and U→S is affine (hence separated)' is inaccurate: the open subscheme U=C∖∪σ_i(S) is not necessarily affine over S. Only separatedness of U→S is used, and it holds because C/S is separated; the wording should be corrected.","section":"§4.2, proof of Proposition 4.23"},{"comment":"The sentence 'We always have (iii) for flat maps T→S' is confusing, since (iii) is part of the proposition being proved. Please clarify that a flat-base-change version of (iii) follows from Proposition 4.18 and Theorem 4.16 and can be invoked at that point.","section":"§4.4, proof of Proposition 4.19(i)"},{"comment":"In Theorem 1.1(ii) the two schemes C×_S T and C×_S T are typeset identically, making the statement hard to parse. Please use distinct notation for the contracted curve in the displayed isomorphism.","section":"Theorem 1.1(ii)"},{"comment":"The name 'Battistella and Carrocci' appears in the Introduction while the reference list has 'Battistella and Carocci'; the spelling should be made consistent.","section":"Introduction and references"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious and detailed contribution, but the false Lemma 4.15 is a genuine obstruction to the proof as written. The rest of the argument—in particular the singularity description in §4.6 and the properness argument in §4.7—appears carefully executed, and the gap is likely repairable by replacing Lemma 4.15/Theorem 4.16 with a correct cohomology-and-base-change argument or a citation of the standard theorem. I would be willing to review a revised version. The paper's own admission in §1.1 about the genus-two limitation should also be reflected in the abstract to avoid scope overreach."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper before choosing whether to spend time on it. First, the main theorem is real: for a family of log curves equipped with a 'mesa' section λ, it constructs a contraction of the subcurve E=|λ| that commutes with arbitrary base change, and it gives explicit control over the resulting singularities. Second, the paper is honest about what it does not do: the construction needs H^1(E,O_E(−λ))=0, and Section 1.1 concedes that this fails for the topological models of Battistella's genus-two Gorenstein singularities. So the scope is narrower than the introduction's 'any genus' phrasing might suggest, but that is a boundary of the method, not a hidden flaw.\n\nWhat is actually new is the mesa curve formalism and the direct construction of the contracted curve from a structure sheaf that is visibly compatible with base change. The prior contractions of RSW, Parker, and HH09 are recovered as special cases, and the explicit ring description near the singularity (Proposition 4.28) is useful and convincing. The proof is long but the architecture is sound: Proposition 4.18 turns acyclicity into the higher cohomology vanishing that drives flatness and base change, and Proposition 4.19 assembles those ingredients cleanly. The examples in 4.6—cusps glued transversally, tacnodes, and the variation of log structure in the tacnode family—are concrete and help the reader see what the abstract construction does.\n\nThe soft spots are proportionate. Most importantly, the acyclicity assumption is load-bearing: without it, Proposition 4.18 collapses, and with it, the genus-two Gorenstein contractions of Battistella are out of reach. The author says this explicitly, so a referee should not treat it as a hidden assumption, but the introduction should probably lead with the boundary rather than the promise. The omitted proof of Lemma 4.15 is a minor but real gap; it is a standard truncation of a complex of flat modules, and the author should supply it in revision. There is no circularity and no parameter fitting; the cited special cases are used as comparisons, not as premises.\n\nWho is this for? Anyone working on modular compactifications of M_g,n, log geometry, or contraction constructions for families of curves. It deserves a serious referee: the central argument holds up, the contribution is new, and the limitations are stated in the paper itself. Send it to review; ask for the Lemma 4.15 proof and a slightly more candid introduction.","headline":"A genuinely new base-change-compatible contraction for log curves, built on a clearly stated acyclicity hypothesis that also marks the honest boundary of the method.","tokens_in":38264,"tokens_out":1265,"would_cite":true,"duration_ms":16200,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14H10","14H20"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every mesa curve admits a contraction of its supported subcurve that is compatible with base change and yields controlled singularities, including elliptic Gorenstein singularities.","keywords":["logarithmic geometry","contraction maps","moduli of curves","mesa curves","elliptic Gorenstein singularities","tropical curves","base change","nodal curves"],"falsifier":"Take the genus-two Gorenstein singularity models discussed in the paper (the topological semistable models that fail the acyclicity condition), compute $H^1(E, O_E(-\\lambda))$ for the corresponding $\\lambda$, and check whether the explicitly defined B-ring is flat over the base or whether the contraction maps fail to commute with specialization; a nonvanishing $H^1$ that yields non-flatness would show the acyclicity assumption cannot be dropped.","tokens_in":37269,"feed_emoji":"📐","tokens_out":13068,"duration_ms":109426,"temperature":0.7,"pith_summary":"The paper constructs a general contraction map for families of log curves: given a \"mesa curve\" (a log curve with a piecewise linear function $\\lambda$ on its tropicalization, the dual graph with edge lengths, defining a subcurve $E$), it collapses $E$ in each fiber to a singularity, preserving arithmetic genus, and does so in a way that commutes with arbitrary base change. Earlier contraction constructions relied on choosing a line bundle and taking the projective spectrum of its pushforwards, which does not commute with base change; this paper instead explicitly writes down the structure sheaf of the contracted curve, gaining base-change compatibility and control over the singularities. The resulting singularities include the elliptic Gorenstein singularities (for steep genus-one mesas), and the construction generalizes the centrally aligned curves used in genus-one moduli work. The load-bearing hypothesis is an acyclicity condition $H^1(E, O_E(-\\lambda)) = 0$.","feed_headline":"Contraction of log-curve subcurves now commutes with base change","feed_subtitle":"It writes down the contracted curve's functions explicitly, so specializations are predictable and moduli-space maps follow.","key_machinery":"The central object is the mesa curve: a proper log curve $\\pi: C \\to S$ over a fine-and-saturated (fs) log scheme together with a section $\\lambda$ of the characteristic sheaf (the quotient of the log structure by units) whose associated piecewise linear function on the tropicalization has support $E = |\\lambda|$ of positive genus, is constant on the core, has slopes $0$ or $-1$ on paths from the core to the complement, and satisfies the acyclicity condition $H^1(E, O_E(-\\lambda)) = 0$. The key mechanism is the B-ring $B(U) = \\Gamma(U, O_C(-\\lambda)) \\oplus \\Gamma(S, O_S)$ with multiplication $(f,c)\\cdot(g,d) = (\\lambda(fg) + df + cg, cd)$, quotiented by the ideal generated by the image of $\\Gamma(S, O_S(-\\rho)) \\to B(U)$; this ring is the structure sheaf of the contracted neighborhood, and its flatness and base-change behavior are controlled by a vanishing theorem for higher cohomology of $O_C(-\\lambda)$ (Theorem 4.16).","core_discovery":"The main theorem (Theorem 1.1) states that for any fine-and-saturated (fs) log scheme $S$ and mesa curve $(\\pi: C \\to S, \\lambda)$, there exists a contraction $\\tau: C \\to \\underline{C}$ of $E = |\\lambda|$ inside $C$ such that each connected component of $E$ that is the support of a steep mesa of genus one contracts to an elliptic Gorenstein singularity, and for every morphism $T \\to S$ the contracted curve pulls back to the contraction of the pullback curve, naturally. The contraction is constructed locally by defining the ring of functions on a neighborhood of the contracted point as $B(U) = \\Gamma(U, O_C(-\\lambda)) \\oplus \\Gamma(S, O_S)$, modulo the image of $\\Gamma(S, O_S(-\\rho))$, with an explicit multiplication; this ring is shown to be flat and its formation to commute with base change, and the contracted curve is obtained as a pushout. The proof verifies the flatness and base-change properties via vanishing of higher cohomology of $O_C(-\\lambda)$, which follows from the acyclicity condition, and then gives an explicit description of the local ring near the singularity: functions on the normalization $Z$ whose boundary values $[f(p_i)]$ lie in a codimension-$g$ subspace $V$ of $k^m$ determined by a Mittag-Leffler problem on $E$.","pith_inferences":["Because the acyclicity condition excludes the genus-two Gorenstein models discussed in the paper, a natural next step is to seek a modified B-ring whose flatness does not require $H^1(E, O_E(-\\lambda)) = 0$; if such a ring exists, the construction might extend to those singularities.","The local-ring description in terms of a codimension-$g$ subspace $V$ suggests the contraction mechanism is really a Mittag-Leffler interpolation condition; this may transfer to other moduli problems where contractions are defined by linear conditions on boundary values, such as weighted stable maps.","Base-change compatibility means the contracted image of an individual curve can be computed directly from the fiber mesa structure, without first building a smoothing family; this could simplify practical checks in moduli computations."],"forward_implications":["If the theorem is right, every mesa curve yields a flat family of singular curves with reduced geometric fibers, and the contraction is compatible with arbitrary base change.","The construction induces morphisms between moduli spaces of curves by contracting the universal curve of a logarithmic moduli space.","Steep genus-one mesas contract specifically to elliptic Gorenstein singularities, so cusps, tacnodes, and transverse unions of them appear as allowed singularities in the target moduli.","The explicit description of the local ring near the contracted point turns the singularity type into a linear condition on boundary values, making the resulting curves amenable to further modular classification."],"supporting_citations":[{"why":"Supplies the local structure theorem for log curves (Theorem 2.3) that underpins the charts and the tropicalization description.","marker":"[Kat00]"},{"why":"Provides the centrally aligned curves that mesa curves generalize, the formula for line bundles $O_C(\\sigma)$ from log sections (Proposition 2.11), and the Proj-based contraction construction this paper improves.","marker":"[RSW17]"},{"why":"Provides the atomic-neighborhood technique and the original genus-one contraction construction behind the standard-situation reduction.","marker":"[Par17]"},{"why":"Gives the characterization of elliptic Gorenstein singularities (Lemma 4.27) used to prove Theorem 1.1(i).","marker":"[Smy11]"},{"why":"Provides the tropical-curve framework of weighted edge contractions and piecewise linear functions used to encode mesa data and describe sections of the characteristic sheaf (Proposition 2.10).","marker":"[CCUW17]"},{"why":"Identifies the genus-two Gorenstein semistable models whose failure of acyclicity delimits the theorem's scope.","marker":"[Bat19]"}],"fun_headline_variants":["Base-change-compatible contractions for log curve subcurves","Log-curve subcurve contractions now commute with base change","Explicit contractions of log-curve subcurves, base-change natural","Contractions of log-curve subcurves induce moduli-space maps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction goes through only when the acyclicity condition $H^1(E, O_E(-\\lambda)) = 0$ holds; if that cohomology group is nonzero for some input, the flatness and base-change compatibility of the contracted family are no longer guaranteed.","fun_headline_variants_meta":{"raw":{"variants":["Base-change-compatible contractions for log curve subcurves","Log-curve subcurve contractions now commute with base change","Explicit contractions of log-curve subcurves, base-change natural","Contractions of log-curve subcurves induce moduli-space maps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001585,"raw_usage":{"total_tokens":6342,"prompt_tokens":984,"completion_tokens":5358,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":5285}},"tokens_in":600,"tokens_out":5358,"duration_ms":37969,"temperature":1.0,"reasoning_tokens":5285,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:03:01.900485+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the genus-two Gorenstein singularity models discussed in the paper (the topological semistable models that fail the acyclicity condition), compute $H^1(E, O_E(-\\lambda))$ for the corresponding $\\lambda$, and check whether the explicitly defined B-ring is flat over the base or whether the contraction maps fail to commute with specialization; a nonvanishing $H^1$ that yields non-flatness would show the acyclicity assumption cannot be dropped.","supporting_citations":[],"review_version":1}