{"id":"b700cddf-3d1d-4c6c-acb7-1b669c827e42","arxiv_id":"2509.25607","paper_version":2,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":2.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A survey of the degeneration theory of Ricci-flat Kahler metrics on Calabi-Yau manifolds: smooth limits for semiample and nef-and-big classes, path-dependent counterexamples at the boundary, and a list of open conjectures.","lead":"This paper surveys what is known about the unique Ricci-flat Kahler metric on a Calabi-Yau manifold as the Kahler class degenerates toward the boundary of the Kahler cone, summarizing theorems and posing open problems. It is the write-up of a research program that recently achieved smooth collapsing asymptotics for fibered Calabi-Yau manifolds and produced counterexamples where the limit depends on the path taken.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified. The survey's load-bearing assertions are published, peer-reviewed results; the in-preparation reference [6] is a minor caveat, not a central dependency.","rationale":"The survey is best read as a position/survey paper, not as a new theorem presentation. The reader's weakest_assumption—the quasi-isometry estimate (14)—is actually a published theorem of [69], not an unproved assumption; surveys routinely rely on peer-reviewed results without reproducing their proofs. That dependency is therefore not a soft spot in the survey itself. I checked for internal inconsistencies and found none: the semiample collapsing results in §3.3 are compatible with the negative examples in §3.4, and the paper explicitly marks its conjectures as open. The one genuine evidentiary gap is the use of the in-preparation reference [6] to support a concrete non-convergence statement in §3.1; this was also flagged by the reader. However, that statement is not load-bearing for the survey's central claims, since Theorem 3.6 already establishes that Question 1.5(b) can fail, and the in-preparation observation is presented as supplementary. Thus no load-bearing concern is identified, and the UNVERDICTED verdict should stand.","tokens_in":21438,"tokens_out":14601,"duration_ms":117036,"concrete_test":"Obtain the in-preparation manuscript [6] (J. Cao) and check whether the asserted non-convergence of CY([β_i]) to η_2 in C^0_loc(X\\W) for every proper closed analytic subvariety W in the §3.1 example is proved. If the manuscript is unavailable or the claim is not verified, revise the sentence to rely instead on Theorem 3.6 as the supporting counterexample.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper is a survey, not a proof of new theorems. Its central assertions—Theorem 2.3, Theorem 3.4, Theorem 3.1, Theorem 3.6—are published results with correct-looking attributions. The reader's identified weakest assumption, the uniform quasi-isometry estimate (14) in §3.3.2, is a theorem of [69], not an unproved assumption; using it as a black box in a survey is standard. The only concrete evidentiary gap is in §3.1, where the claim that CY([β_i])→η_2 fails to converge in C^0_loc(X\\W) for any proper subvariety W is attributed to 'J. Cao, in preparation' ([6]). This is a real missing-support flag, but it is not load-bearing: Theorem 3.6 independently demonstrates failure of C^0_loc convergence (along a ray to a non-semiample class), so the survey's conclusion that Question 1.5(b) can fail does not depend on [6]. No internal inconsistency or circular step found. Verdict UNCHANGED.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper is a survey of the author's and collaborators' work on the behavior of the Calabi-Yau map CY: C → H as a Kähler class approaches the boundary of the Kähler cone. The Calabi-Yau theorem provides a unique Ricci-flat Kähler metric in each Kähler class, and the survey asks whether this map extends continuously to nef classes, what regularity the limiting currents have, and whether convergence holds in stronger topologies away from an exceptional set. The main results surveyed are: Theorem 2.3 (continuous extension and C^∞_loc convergence off the null locus for nef and big classes, attributed to [68,2,8]); Theorem 3.1 (non-uniqueness of weak limits for two sequences approaching the same non-big nef class on a K3 surface); Theorem 3.4 (for semiample classes, weak convergence to f^*(ω_Y + i∂∂φ_0), independence of [ω], and C^∞_loc convergence away from singular fibers, with the refined asymptotic expansion (21) and explicit first nonlinear term (22)); and Theorem 3.6 (a non-semiample example where C^0_loc convergence fails). Section 4 collects results on Gromov-Hausdorff limits, diameter bounds, and the collapse to the metric completion of (Y\\D, ω_0).","tokens_in":21524,"tokens_out":4747,"duration_ms":40050,"significance":"If correct, the surveyed results give a fairly complete picture for semiample classes and for nef-and-big classes, with sharp counterexamples for general nef classes. The paper's main value is as a roadmap: it states precise hypotheses for published theorems, explains the mechanisms behind the higher-order estimates (quasi-isometry (14), stretched PDEs, blow-up/Liouville arguments, shrinking Hölder norms), and isolates open questions (Question 3.5, Conjectures 2.7, 3.7, 3.8, 3.9, 4.2). The central assertions are not new proofs but are attributed to peer-reviewed sources. One concrete gap is the use of the in-preparation reference [6] in §3.1 for the assertion that CY([β_i]) does not converge in C^0_loc(X\\W) for any proper subvariety W; this is not load-bearing because Theorem 3.6 independently provides a published counterexample to Question 1.5(b). The survey would be strengthened by either replacing [6] with a published reference or explicitly noting that the conclusion also follows from Theorem 3.6.","major_comments":[],"minor_comments":[{"comment":"The claim that CY([β_i])→η_2 does not converge in C^0_loc(X\\W) for any proper closed analytic subvariety W is attributed to 'J. Cao, in preparation' ([6]). Since this claim is used to conclude that Question 1.5(b) may fail, please provide the full reference if it now exists, or state that the same conclusion follows from the published Theorem 3.6. This is a local presentation issue, not a load-bearing dependency.","section":"§3.1"},{"comment":"The notation 'R[α]∩H^2(X,Q)' is ambiguous as printed. It presumably denotes the real line ℝ[α] through [α]; please use ℝ[α] or explain the notation, since otherwise the intersection with H^2(X,Q) is hard to parse.","section":"Conjectures 3.9 and 4.2"},{"comment":"The 'shrinking Hölder norms' C^{k,α}(g_t) are used in the displayed asymptotic expansion (21) and in the surrounding text, but they are only described informally. A precise definition or a precise pointer to [40, §2] would make the statement of the expansion more self-contained.","section":"§3.3.6"},{"comment":"Reference [6] appears only as 'J. Cao, in preparation' with no year. If the manuscript is to be published, please update this entry or remove the dependence on it, as discussed above.","section":"§3.1 / References"}],"recommendation":"minor_revision","confidential_remarks":"The survey is heavily self-referential, which is natural for a survey of the author's own program, but the editor may wish to confirm that the venue is comfortable with that format. The mathematical content is sound as far as it goes; the only concrete gap is the in-preparation reference [6], which is not central to the conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"It is a survey, not a research announcement, and judging it as a research paper would be a category error. The real value is that Tosatti has written a precise, honest map of the degeneration theory of Ricci-flat metrics on Calabi-Yau manifolds, centered on the work he and his collaborators have done over the last decade. The statements of the main results — Theorem 2.3 on nef-and-big classes, Theorem 3.4 on smooth collapsing for semiample classes, and the counterexamples in Theorems 3.1 and 3.6 — are accurate and well attributed. The exposition of the Hein–Tosatti asymptotic expansion, including the explicit leading correction term in (22) in terms of Kodaira–Spencer forms, is about as clear as one could hope for in a survey. The genuinely new content is a small set of open problems and conjectures (3.5, 3.7, 3.8, 4.2), and some of these, especially the refined expansion questions in 3.5, are likely to shape the field.\n\nThe soft spots are minor and mostly inherent to the genre. The paper is heavily self-referential, but that reflects the reality that Tosatti (and his coauthors) proved many of these theorems; it is not a flaw as long as a reader cross-checks the primary sources. Two unrefereed references appear in load-bearing positions: [6] (Cao, in preparation) is cited for a claim about failure of C^0 convergence, and [67] (Szekelyhidi) for the latest Gromov-Hausdorff results. Of these, [6] is not essential, because Theorem 3.6 already demonstrates that stronger convergence can fail, and the stress-test note is right on that point. [67] is more central to Theorem 4.6(c), but it is attributed clearly as a recent preprint, which is normal in a survey. The only genuinely uncited assertion I noticed is the folkloric claim in Example 1.2 about tori being essentially the only explicit examples; that is harmless as folklore.\n\nThe survey deserves refereeing: a serious editor should send it out, because accuracy of attribution and statement in a survey that will become a standard reference matters, and because a referee can check the new conjectures for well-posedness. It is not a desk reject. I would bring it to a reading group for anyone entering the collapsing Calabi-Yau literature, and I would cite it for the clean statements and the list of open problems.","headline":"A careful, useful survey of a fast-moving area from the person who proved many of the key results; not a new-theorem paper, but a solid map of the territory and a set of fresh conjectures worth refereeing.","tokens_in":714,"tokens_out":974,"would_cite":true,"duration_ms":21173,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C25","53C55","32Q25","32W20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For semiample classes, collapsing Calabi-Yau metrics admit a complete asymptotic expansion.","keywords":["Ricci-flat metrics","Calabi-Yau manifolds","Kähler cone","nef classes","semiample classes","collapsing","asymptotic expansion","Kodaira-Spencer forms"],"falsifier":"Compute the asymptotic expansion (21) on a specific nontrivial elliptic fibration of a K3 surface and check whether the first correction term has the predicted t^2 scaling with the explicit coefficient from (22); any discrepancy, or a semiample fibration where the C^∞ convergence away from the singular fibers fails, would refute Theorem 3.4. Alternatively, find a sequence of Kähler classes converging to an irrational nef boundary class on a K3 surface whose diameter stays bounded away from zero; this would disprove Conjecture 4.2.","tokens_in":21116,"feed_emoji":"🌀","tokens_out":7435,"duration_ms":51336,"temperature":0.7,"pith_summary":"This paper surveys the space of Ricci-flat Kähler metrics on a fixed Calabi-Yau manifold and asks what happens to the unique Ricci-flat metric as its Kähler class approaches the boundary of the Kähler cone. The central picture it presents is that the answer is exactly determined by the type of the limiting class: for nef and big classes the Calabi-Yau map extends continuously and the metrics converge smoothly away from a null locus; for semiample classes the collapsing metrics converge smoothly away from the singular fibers and admit an explicit asymptotic expansion with a leading correction built from the Kodaira-Spencer forms; and for general nef classes no continuous extension exists, since different paths to the same boundary class can give different weak limits. The paper also lays out a series of conjectures separating the understood semiample territory from the open general case. A reader should care because these degenerations are the canonical geometric objects attached to a Calabi-Yau manifold, and the explicit expansion turns a qualitative collapse into a computable one.","feed_headline":"Calabi-Yau metrics collapse smoothly for semiample classes","feed_subtitle":"General boundary classes can have several weak limits; an explicit first-order correction is known.","key_machinery":"The central object is the Calabi-Yau map, which assigns to each Kähler class its unique Ricci-flat metric, and its extension to the boundary of the Kähler cone as a map into closed positive currents. For the semiample collapse, the load-bearing mechanism is the uniform quasi-isometry estimate (14), which compares the collapsing metrics ω_t to the model shrinking metrics f^*ω_Y + tω on compact sets away from the singular fibers. This allows a stretching argument that makes the degenerate Monge-Ampère equations uniformly elliptic, yielding higher-order estimates. The asymptotic expansion is organized using a fiberwise connection D (which acts as the Levi-Civita connection of each fiberwise Ric","core_discovery":"The paper's central claim is that the degeneration behavior of Ricci-flat Kähler metrics is governed by the position of the limiting class in the Kähler cone. On nef-and-big classes, the Calabi-Yau map extends continuously and injectively, and Ricci-flat metrics in approaching Kähler classes converge in C^∞ locally away from the null locus. When the limit class is semiample, the metrics along the ray [α]+t[ω] converge weakly to a pullback metric and admit a detailed asymptotic expansion, ω_t = η + t ω_SRF + i∂∂ψ_t + Σ_{j=2}^k γ_{j,k} + η_k, where the first nontrivial term γ_{2,k} is explicitly given by inverting the fiberwise Laplacian applied to the Kodaira-Spencer forms. In contrast, for g","pith_inferences":["The explicit form of γ_{2,k} suggests a practical test: computing the first correction to the fiberwise volume or to the total scalar curvature in an elliptic K3 fibration should show a characteristic t^2 term governed by the variation of complex structure; a numerical check on a known family would confirm the expansion's coefficients.","Because the continuous extension fails for general nef classes, any construction of canonical 'large complex structure limits' must either restrict to semiample or big classes or specify a choice of path in the Kähler cone; the path-dependence is a geometric counterpart of the multi-valuedness of limiting objects.","The connection D and shrinking Hölder norms introduced for the expansion likely form a reusable template for other degenerate Monge-Ampère problems, such as the Kähler-Ricci flow collapsed limits mentioned in the survey, where the same techniques have already resolved long-standing conjectures.","The conjectures about irrational nef classes on K3 surfaces suggest that dynamics (automorphisms with positive entropy) might force uniqueness of currents; if true, this would add a new bridge between complex dynamics and metric degeneration."],"forward_implications":["For semiample classes, the Ricci-flat metrics collapse smoothly to a Kähler metric on the base of the fibration, away from the singular fibers, and the first-order correction is computable from the fiberwise geometry.","The Gromov-Hausdorff limit of the collapsing metrics is the metric completion of the base with the limiting metric, regular outside a set of real Hausdorff codimension at least 2 (and 4 in the semiample case).","The Calabi-Yau map cannot extend continuously to all nef boundary classes: there exist K3 surfaces with two sequences of Kähler classes converging to the same boundary class whose Ricci-flat metrics converge to different currents.","Weak limits of Ricci-flat metrics along nef-but-not-big classes are not in general continuous, so any bounded-potential regularity must be proved class-by-class.","If the conjectures on bounded potentials and minimal singularities hold, the uniform L^∞ estimate (24) would settle the general regularity question for weak limits."],"fun_headline_variants":["Semiample classes: smooth collapse with explicit first-order expansion","Boundary classes have multiple weak limits; semiample collapse smoothly","Explicit first-order correction for collapsing Calabi-Yau metrics"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The semiample collapse theorem rests on the uniform quasi-isometry estimate (14), which says that on any compact set avoiding the singular fibers, the collapsing Ricci-flat metrics are uniformly comparable to the model metrics f^*ω_Y + tω; if this estimate fails for some fibration, the higher-order estimates and the asymptotic expansion (21) do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Semiample classes: smooth collapse with explicit first-order expansion","Boundary classes have multiple weak limits; semiample collapse smoothly","Explicit first-order correction for collapsing Calabi-Yau metrics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000907,"raw_usage":{"total_tokens":3651,"prompt_tokens":575,"completion_tokens":3076,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":319,"completion_tokens_details":{"reasoning_tokens":3019}},"tokens_in":319,"tokens_out":3076,"duration_ms":35900,"temperature":1.0,"reasoning_tokens":3019,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T13:42:38.601374+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the asymptotic expansion (21) on a specific nontrivial elliptic fibration of a K3 surface and check whether the first correction term has the predicted t^2 scaling with the explicit coefficient from (22); any discrepancy, or a semiample fibration where the C^∞ convergence away from the singular fibers fails, would refute Theorem 3.4. Alternatively, find a sequence of Kähler classes converging to an irrational nef boundary class on a K3 surface whose diameter stays bounded away from zero; this would disprove Conjecture 4.2.","supporting_citations":[],"review_version":1}