{"id":"33457db1-eb57-4ef6-9e03-d237172dbd25","arxiv_id":"2411.10157","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Symplectic non-Hamiltonian circle actions on compact connected 4-manifolds with discrete period group are classified up to equivariant symplectomorphism by seven invariants.","lead":"A new classification theorem describes all symplectic non-Hamiltonian circle actions on four-dimensional manifolds, up to natural equivalence, using seven invariants. This extends the classical classifications of Hamiltonian actions and gives explicit tools for understanding Kähler structures and further symmetries.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 7's sheaf-cohomology isomorphism is asserted by transfer from [KT03] without verifying the local-isotopy lemmas over the circle R/P; the uniqueness theorem depends on it.","rationale":"The reader's weakest assumption was the discreteness of the group of periods, which is an explicitly stated scope limitation rather than a hidden flaw. My stress-test identifies a more specific load-bearing spot: the transfer of the Karshon–Tolman sheaf-theoretic classification from intervals to the circle R/P in Section 7. This transfer is asserted rather than proved, and the classification theorems (Proposition 8.9 and hence Theorem 2.9) depend on it. The concern does not constitute a demonstrated counterexample, but it is a precise place where the argument could break. The reader's CONDITIONAL verdict already accounts for deferred proofs; my concern sharpens that condition by naming the exact deferred step and offering a concrete verification. I therefore recommend no change to the verdict: the paper should be accepted only after the sheaf-cohomology isomorphism over the circle is fully verified, and the proposed direct computation would serve as a meaningful check of that step.","tokens_in":58708,"tokens_out":21851,"duration_ms":217436,"concrete_test":"Work out the sheaf cohomology groups ˇH^1(R/P, Q_{R/P}) and ˇH^1(R/P, P_{R/P}) directly for the simplest nontrivial case: free actions with genus 1 and trivial monodromy (Example A.3), using a good cover of the circle by three open intervals and explicit transition maps for the trivial torus bundle. Compare the resulting Φ-diffeomorphism classes with those predicted by the monodromy invariant and the fibration invariant in the appendix. If the direct computation yields additional classes not distinguished by the monodromy invariant, then Corollary 7.18 is false in the circle-valued setting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The classification up to equivariant symplectomorphism (Theorem 2.9) depends on Proposition 8.9, which classifies spaces up to Φ-diffeomorphisms using the genus, isotropy data, and monodromy invariants. Proposition 8.9 in turn relies on Corollary 7.18, the isomorphism ˇH^1(R/P, Q_{R/P}) ≅ ˇH^1(R/P, P_{R/P}) between the sheaf of circle-valued Hamiltonian S1-pieces and the sheaf of legal painted surface bundles. The proof of this isomorphism is given by Lemma 7.11, which asserts a chain of isomorphisms ˇH^1(ˆQ) ≅ ˇH^1(RQ) ≅ ˇH^1(E) ≅ ˇH^1(RP) ≅ ˇH^1(ˆP) by saying that the objects 'can be regarded as proper tall grommeted complexity one spaces' and therefore the propositions of [KT03] 'also apply to our case.' This transfer is not automatic: the sheaves in [KT03] are defined over an interval (the momentum image of a complexity one space), while here the base is a circle R/P. The local isotopy lemmas from [KT03, Props 13.19, 14.9, 14.12, 15.10], used to verify the hypotheses of Lemma 7.12, may rely on the contractibility of the base in a way that fails for a circle. In particular, Lemma 7.12 involves finite unions W of open sets in a cover, and the verification of condition (3) could be more delicate when intersections of cover elements have two components. If this isomorphism is not valid in the circle-valued setting, then Proposition 7.20 and hence Proposition 8.9 are unsupported, and the completeness of the seven invariants in Theorem 2.9 is not established. This is a load-bearing gap because the rest of the paper's uniqueness proof assumes this bridge between geometry and topology.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines seven invariants for compact connected symplectic 4-manifolds equipped with symplectic non-Hamiltonian circle actions whose group of periods is discrete: the group of periods, the Duistermaat-Heckman constant, the genus invariant, the isotropy data invariant, the monodromy invariant, the fibration invariant, and the de Rham invariant. It then proves an existence theorem (Theorem 2.8) realizing every valid combination of these invariants and a uniqueness theorem (Theorem 2.9) stating that two such spaces are equivariantly symplectomorphic if and only if the seven invariants agree. The main strategy is to choose a tight circle-valued Hamiltonian, reduce the classification to painted surface bundles over R/P via the sheaf-theoretic machinery of Karshon-Tolman, and then classify those bundles by the monodromy, fibration, and de Rham data.","tokens_in":59000,"tokens_out":3486,"duration_ms":37118,"significance":"If the main theorems are correct, the paper gives a complete solution to a natural extension of Karshon's classification of Hamiltonian circle actions and Pelayo's classification of symplectic T^2 actions, under a discreteness assumption that covers rational symplectic forms and quotients with first Betti number one. The paper is original and technically ambitious: it proves a fiber connectivity lemma for circle-valued Hamiltonians, establishes constancy of the Duistermaat-Heckman measure, develops a monodromy invariant capturing braiding of non-free orbits, and gives explicit computations in Appendix A. The self-contained proof in Appendix B that smooth and topological mapping class groups agree for surfaces with marked points is also a useful contribution. The main unresolved issue is whether the transfer of the Karshon-Tolman sheaf-theoretic classification from an interval base to the circle base R/P is justified; this is load-bearing for the uniqueness theorem.","major_comments":[{"comment":"The proof of Lemma 7.11 asserts a chain of isomorphisms ˇH^1(R/P, Q_{R/P}) ≅ ˇH^1(R/P, P_{R/P}) by saying that the objects can be regarded as proper tall grommeted complexity one spaces and therefore the propositions of [KT03] apply. This transfer is not automatic. In [KT03] the base of the sheaves is the momentum image, which is an interval, whereas here the base is the circle R/P. The local isotopy results [KT03, Props 13.19, 14.9, 14.12, 15.10] are used to verify the hypotheses of Lemma 7.12, and the verification of condition (3) in Lemma 7.12 involves finite unions W of open sets and may be more delicate when intersections of cover elements have two connected components. The paper gives no argument addressing this difference. Since Corollary 7.18, Proposition 7.20, Proposition 8.9, and ultimately Theorem 2.9 all depend on this isomorphism, this is a load-bearing gap that must be repaired.","section":"§7, Lemma 7.11"},{"comment":"The proof of Proposition 7.20 is delegated to the statement that it is identical to the proof of Proposition 17.5 in [KT03]. That proposition is proved for proper tall complexity one spaces over an interval, but here the associated painted surface bundle lives over the circle R/P. The existence of a global object and the homeomorphism Ψ : M/S^1 → N require gluing over the two ends of the interval, and this is precisely where the circle-valued setting differs from the Hamiltonian setting. The author should either supply the full argument or state a precise reduction showing that the circle base causes no change to the proof.","section":"§7, Proposition 7.20"},{"comment":"Lemma 9.6 is a strengthened version of the local uniqueness theorem [KT01, Theorem 1] and is used in Theorem 9.1 to realize an arbitrary prescribed monodromy class α. The proof is only a sketch. In particular, the sentence 'By applying Lemma B.11 to each labeled point, we get a rigid map f' : Σ → Σ′' is not immediate: Lemma B.11 produces an isotopy of a diffeomorphism to the identity near a fixed point, whereas a rigid map must rotate the slice by the prescribed rotation. This step needs a detailed justification, because the construction of the final gluing map Ψ and hence the claimed realization of the monodromy invariant depends on it.","section":"§9, Lemma 9.6"}],"minor_comments":[{"comment":"The phrase 'Dusitermaat-Heckman invariant' should be 'Duistermaat-Heckman invariant'.","section":"§9, first paragraph"},{"comment":"The phrase 'non-trivial principle S1 bundle' should be 'non-trivial principal S1 bundle'.","section":"§2, Remark 2.14"},{"comment":"The statement contains the condition 'X ∩ Y =,' which appears to have a missing symbol, likely ∅. As written it is difficult to parse.","section":"§7, Lemma 7.12"},{"comment":"The computation of the de Rham invariant action would benefit from a displayed formula showing how the SL2(Z) matrix acts on (m,n), rather than only describing it in prose.","section":"Appendix A.3"},{"comment":"The phrase 'there are infinitely many pairs of non symplectomorphic spaces in this family' should be 'infinitely many isomorphism classes' or similar, since 'pairs' is not the intended meaning.","section":"§5, Example 5.11"}],"recommendation":"major_revision","confidential_remarks":"The paper makes a strong claim and the overall architecture is plausible, but the §7 transfer from [KT03] to the circle base is the key technical risk. If the author can provide a detailed verification of Lemma 7.11 and Proposition 7.20 in the circle-valued setting, the classification would be convincing. The paper is not relying on any circular reasoning or fitted parameters; it builds on published work of Karshon-Tolman and McDuff. The main issue is completeness of the proof, not novelty."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a substantial paper and probably the right answer to Problem 1.3 under the discreteness assumption. The seven invariants are genuinely new, and the monodromy invariant capturing braided non-free orbits is a real addition, not a repackaging of Karshon–Tolman's painting invariant. The paper is also honest: the discreteness assumption is stated up front, the non-discrete case is explicitly left open (Remark 2.11), and the appendix computes actual invariants for concrete examples rather than just asserting they exist.\n\nThe main theorems are independent of the prior classifications; they build on McDuff and Karshon–Tolman but do not reduce to them. The analytic lemmas that carry the paper — fiber connectivity and constant Duistermaat–Heckman measure — are proved in detail. The construction of spaces from invariants is concrete and uses a strengthened local uniqueness statement that is at least plausible from the KT01 proof.\n\nThe soft spot is Section 7. The isomorphism between the sheaf of circle-valued Hamiltonian pieces and the sheaf of painted surface bundles (Corollary 7.18) is asserted by transfer from [KT03], and the transfer is not automatic because the base is a circle rather than an interval. The local isotopy lemmas in [KT03] are local, so I suspect the adaptation goes through, but Lemma 7.11 needs a real check, especially for covers of the circle whose intersections have two components. This is load-bearing for the uniqueness theorem. It is not obviously fatal, but a referee should ask for the transfer to be proved or for precise references to the exact propositions with an explanation of why the circle base does not matter.\n\nThere are smaller terseness issues: several \"proof is identical to [KT03]\" claims are fine, but Proposition 7.20 is important enough to deserve a sketch. None of this looks like circularity or fitted parameters; the invariants are intrinsic, and the proof structure is coherent.\n\nI agree with the conditional verdict. The paper deserves a serious referee. If the Section 7 gap is fixed — and I expect it can be — this should go in a strong journal. I would send it to review and ask for the transfer argument to be written out.","headline":"A serious classification theorem for non-Hamiltonian circle actions in dimension four, with genuinely new invariants; the main soft spot is a terse transfer from Karshon–Tolman that a referee should ask to be expanded.","tokens_in":59663,"tokens_out":2542,"would_cite":true,"duration_ms":30318,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D20","57R17"],"pacs":[],"model":"deepseek-v4-flash","headline":"Seven invariants completely classify symplectic non-Hamiltonian circle actions on 4-manifolds under a discreteness assumption.","keywords":["symplectic circle actions","non-Hamiltonian actions","equivariant classification","circle-valued Hamiltonian","surface bundles over a circle","mapping class group","Duistermaat–Heckman theory","Seifert fibrations"],"falsifier":"A concrete counterexample would settle the completeness claim: find two compact connected symplectic 4-manifolds with non-Hamiltonian circle actions and discrete period groups that agree on all seven invariants yet are not equivariantly symplectomorphic. Alternatively, take the torus example with period group $\\mathbb{Z}+\\sqrt{2}\\mathbb{Z}$ and perturb the symplectic form in two different ways to rational classes; if the five topologically defined invariants of the perturbed spaces differ, then the classification's conclusions for non-discrete periods are perturbation-dependent, which the paper itself flags as a possibility.","tokens_in":58389,"feed_emoji":"🔄","tokens_out":12562,"duration_ms":113502,"temperature":0.7,"pith_summary":"This paper solves a classification problem for symplectic actions that are not Hamiltonian: it provides a complete list of invariants that decide when two symplectic non-Hamiltonian circle actions on compact connected symplectic 4-manifolds are equivariantly symplectomorphic. The list comprises the group of periods of $\\iota_X\\omega$, the constant Duistermaat–Heckman density, the genus of the reduced surfaces, the isotropy data of a level set, a monodromy class recording how non-free orbits braid around the circle of momentum values, and two cohomological invariants that control the $S^1$-bundle and the symplectic form. The classification requires the group of periods to be discrete, which the paper shows holds automatically for rational symplectic forms and for quotients with first Betti number one. If correct, it turns the equivalence problem into seven checkable data and exposes new topology, the braiding of non-free orbits, that cannot appear in the Hamiltonian case.","feed_headline":"Seven invariants classify symplectic circle actions on 4-manifolds","feed_subtitle":"Two such 4-manifolds are the same exactly when these seven data agree.","key_machinery":"The load-bearing object is the tight circle-valued Hamiltonian $\\Phi: M \\to \\mathbb{R}/P$, whose level sets are connected precisely because the group of periods $P(\\iota_X\\omega)$ is discrete. Associated to it is a painted surface bundle over the circle $\\mathbb{R}/P$: a surface bundle in which the non-free orbits are marked and labeled by their isotropy data. A sheaf-of-groupoids comparison shows that two spaces have $\\Phi$-diffeomorphic quotients exactly when their associated painted surface bundles are isomorphic, and those bundles are then classified by the genus, the labeling, and the conjugacy class of the gluing map in the mapping class group $\\mathrm{MCG}(\\Sigma_{g,C_1,\\ldots,C_k})$ of the labeled surface. Two further cohomological devices, the fibration invariant (an orbit of $H^2(M/S^1,\\mathbb{Z})$ under $\\Phi$-diffeomorphisms) and the de Rham invariant (an orbit in basic cohomology controlling the symplectic class), upgrade $\\Phi$-diffeomorphisms first to equivariant diffeomorphisms and then to symplectomorphisms via a Moser-type argument.","core_discovery":"The paper's central claim is that the classification of symplectic non-Hamiltonian circle actions on compact connected symplectic 4-manifolds reduces to seven computable invariants. Theorem 2.9 asserts that, when the group of periods of $\\iota_X\\omega$ is discrete, two such actions are equivariantly symplectomorphic if and only if the two spaces have the same group of periods, Duistermaat–Heckman constant, genus invariant, isotropy data invariant, monodromy invariant, fibration invariant, and de Rham invariant. Theorem 2.8 asserts that every combination of values satisfying the evident constraints is realized by some space, so the list is exhaustive. The geometric picture that makes the completeness proof work is that the quotient $M/S^1$ is a surface bundle over the circle $\\mathbb{R}/P$ whose fibers are the reduced surfaces and whose non-free orbits appear as finitely many labeled points that can braid around the base direction; this braiding is captured by the monodromy invariant, a conjugacy class in the mapping class group of the labeled surface. This topological information is new in dimension four, because the analogous painting invariant for Hamiltonian complexity-one spaces is known to be trivial there.","pith_inferences":["The discreteness hypothesis may be genuinely necessary: the paper's Example 4.5 shows that when the periods are irrational the target $\\mathbb{R}/P$ is not a manifold, and Remark 2.11 leaves open whether the invariants obtained from different perturbations agree; a concrete failure there would require new invariants for the non-discrete class.","The new monodromy invariant is a natural candidate for an obstruction: the paper raises, without resolving, whether it controls the existence of equivariant Kähler or complex structures and whether the circle action extends to a symplectic $T^2$ action.","The same strategy plausibly extends to higher-dimensional complexity-one actions with isotropic orbits, as the paper itself notes; the connectivity lemma (Lemma 2.2) is dimension-free, so the main obstacles are technical rather than conceptual.","Because the fibration and de Rham invariants are orbits of finite-dimensional cohomology actions, computing them for explicit examples (products of surfaces, mapping tori of the torus) is a finite linear algebra problem, even though describing the orbits in general is complicated."],"forward_implications":["Under the discreteness assumption, two spaces are equivariantly symplectomorphic if and only if all seven invariants agree, so the classification problem is solved and the data are checkable in practice (Theorem 2.9).","Every combination of values satisfying the relevant integrality and orbit constraints is realized by some constructed space, so the list is exhaustive (Theorem 2.8).","When the quotient space has first Betti number one, the fibration and de Rham invariants are trivial, so the first five invariants already determine the space (Remarks 10.5 and 11.6).","Every space with non-discrete periods is still equivariantly diffeomorphic to one of the constructed spaces after perturbing the symplectic form, although the invariants of the perturbed space are not shown to be independent of the perturbation (Corollary 2.10 and Remark 2.11).","The quotient of any such space is a surface bundle over a circle with finitely many labeled braids of non-free orbits, providing a concrete topological model for the action (Proposition 2.4)."],"supporting_citations":[{"why":"Introduces circle-valued Hamiltonians and the perturbation argument, and proves the no-fixed-point lemma (Lemma 3.2) for four-dimensional non-Hamiltonian circle actions.","marker":"[McD88]"},{"why":"Provides the local normal form, gluing, and uniqueness techniques for complexity-one spaces that the paper adapts (its Theorem 1 underlies Lemma 9.5).","marker":"[KT01]"},{"why":"Supplies the sheaf-of-groupoids and painted-surface-bundle machinery that reduces the classification to differential topology.","marker":"[KT03]"},{"why":"Extends the classification tools used for the local existence lemma (Lemma 9.2).","marker":"[KT14]"},{"why":"Classifies orbifold $S^1$-bundles via cohomology, giving the criterion for lifting $\\Phi$-diffeomorphisms to $\\Phi$-$T$-diffeomorphisms used in Section 6.","marker":"[HS91]"},{"why":"Yields the Duistermaat–Heckman theorem used to show the density is constant and the Seifert Euler number of level sets is zero (Lemma 2.3).","marker":"[DH82]"},{"why":"Fixes the notation and definition of Seifert invariants and the Seifert Euler number, which encode the isotropy data invariant.","marker":"[JN83]"}],"fun_headline_variants":["Seven invariants settle symplectic circle action classification","Four-manifold symplectic actions pinned by seven invariants","Seven invariants nail symplectic circle actions on 4-manifolds","New monodromy invariant completes symplectic 4D classification"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification assumes throughout that the group of periods of $\\iota_X\\omega$ is a discrete subgroup of $\\mathbb{R}$; when the periods are dense, the space $\\mathbb{R}/P$ is not a circle and the machinery collapses.","fun_headline_variants_meta":{"raw":{"variants":["Seven invariants settle symplectic circle action classification","Four-manifold symplectic actions pinned by seven invariants","Seven invariants nail symplectic circle actions on 4-manifolds","New monodromy invariant completes symplectic 4D classification"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000806,"raw_usage":{"total_tokens":3528,"prompt_tokens":919,"completion_tokens":2609,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":535,"completion_tokens_details":{"reasoning_tokens":2534}},"tokens_in":535,"tokens_out":2609,"duration_ms":22363,"temperature":1.0,"reasoning_tokens":2534,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:55:05.088935+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete counterexample would settle the completeness claim: find two compact connected symplectic 4-manifolds with non-Hamiltonian circle actions and discrete period groups that agree on all seven invariants yet are not equivariantly symplectomorphic. Alternatively, take the torus example with period group $\\mathbb{Z}+\\sqrt{2}\\mathbb{Z}$ and perturb the symplectic form in two different ways to rational classes; if the five topologically defined invariants of the perturbed spaces differ, then the classification's conclusions for non-discrete periods are perturbation-dependent, which the paper itself flags as a possibility.","supporting_citations":[],"review_version":1}